Networking For Cloud Computing | AWS Networking Training | What Is CIDR | Subnetting | VLSM

[Music] okay welcome everyone this is michael gibbs for from go cloud architects how are you all today in any case i'm super happy and super excited to be here with you today today we're going to be talking about one of my favorite topics in the entire world networking so for those of you that don't know me my name is michael gibbs and i'm the founder and ceo of go cloud architects and we're an organization that's really dedicated towards building high performance cloud computing and networking careers the network is the heart of everything it is the foundation it is the plumbing it is the basis of everything that we do and if the network doesn't work right nothing works right and when the network works great everything works great and you know the best part of the network is no one even thinks that it's there so when you've got some network skills you're really in a great place because you're one of the few people that actually knows how it all works so because of that we're going to talk about networking i've been involved in the networking field for a really long time i went from practicing internal medicine to senior network engineer to lead architect in about six months when i began my tech career transition 25 years ago and during this time i've been working and networking forever i got to work for companies like comcast as a principal architect i got to work at cisco as an enterprise architect for almost a decade and what a great experience it was i got to work designing systems for banks and it was great and i love it and networking is the foundation and i want to give you that knowledge and that's why we're here now the specific part of networking we're going to talk about today is subnetting and while everybody thinks of subnetting as the foundation and the basic basic most elemental is part of networking which it is it's also one of the most critical parts of networking in fact where people make errors in networking the people that do the network design and the ip address design should be the most senior people out there why the way the addresses are laid out and the way the addresses are planned will determine if the systems will work now or will work ever or whether they're going to run into problems as the systems grow and scale so networking specifically the addressing plan will determine how everything works on the network and we're going to spend a lot of time on that so we'll talk about subnetting today which is basically taking one set of blocks and breaking it down we'll talk about classical addresses which we don't use anymore so we can get into classes intradomain routing then we'll talk about what goes into subnetting then we'll talk about how to subnet then we'll actually go the opposite direction and in the opposite direction what we're going to do is we're going to talk about how to supernet which is used for route aggregation and route summary then we'll do some practice things and we'll work through it together and then if you guys want we can add some architecture to the mix because we know i love architecture and if you guys want to add some architecture to the mix we can design an enterprise with their subnetting and if you want we can even put in a routing protocol or two and i can teach you how you would design a system maybe with a core access distribution if you want to do some network architecture in there we'll have some fun but we can't do any network architecture and quite frankly we can't do any cloud architecture unless we've got the ip addressing down so that's what we're going to talk about today we're going to have some fun and it's going to be there so if you guys are having fun and want to do some tech and want to do some architecture please in the chat box do this type networking for cloud computing i'm hoping to see a good 50 of them because there's 46 of you on here right now and also if you have if you're ready to have fun if you can hit the like button if you're not a member if you can subscribe if you're a subscriber if you can hit that bell you'll be notified because we've got a lot more free training initiatives coming out for you so networking for cloud computing i'll know that you're here so hello everyone and welcome and while you're popping that in there i can see actually who's coming abigail mark's cloud hired i'm so happy to see you here marla i'm doing great i hope you are too pierre so good to see you hi everyone leo so good to see you speak to you yousef so happy to see you leonard lash i'm doing great and my cat cindy's doing great today it's been a very sweet little grow melon glad hired see you leo hello all nitro pen great seeing you and noel it's always good to see you there q welcome hello chad hi chad perez the devs love some architecture we'll do some architecture alonzo my really good friend um who i called james bond cloudheart good to see you and perez not working for cloud computing david welcome marla welcome and daniel i'm so happy to see her daniel so good to see her today let's get everybody cloud hired so networking for cloud commuting super happy to have you here so let's talk about ip addressing and what it is and how it works and why all of this stuff matters so much pal wonder wonderful carlos wonderful so happy to see you yusuf it makes me so happy to see so many people coming from all around the world io i know a lot of ios but i am super happy to see you here um regardless of which ir you are or if you're another io i'm still thrilled to see you here christina lee uh welcome welcome welcome so let's get in to sudden hitting and let's before we do this let's talk a little briefly about ip addresses now every device in the entire world if it's going to be reachable has to have an address and the addresses are all going to have to be unique so you know you've got a phone somebody wants to call go cloud architects they call 973-896-4889 and then either i'll answer it or the phone will be forwarded to one of my executive team or somebody else will hold the phone and how does the phone company know to send that call to that plus one nine seven three eight nine six four eight eight nine how does the world know it it knows it because the companies the phone companies have switches and the switches basically have a map of basically the network and the switches know where to send the calls to well guess what when we're dealing with computers and we're dealing with network communication here's what happens we've got these devices called routers and routers actually build a map of the network and what goes on is they know how to get from point a to point b and the route will route your traffic through the network but what are the routers actually looking at what is the routers building a map of they're building a map of all the subnets along the way and which interfaces to use to actually get to which subnet so every device is going to have to have its own unique address and every device that's on a subnet can only talk to devices on that subnet or on that network to go so to go from one subnet and we'll talk about what these subnets are in things later to another subnet or a class b address to a different class b in a different network you need a router so that's the key element to all of this thing is what is networking networking is basically connecting devices networking provides the plumbing every device needs a unique address and the address has to be on the same network as the other devices that it's talking about and when we talk about the same network we're actually talking about the same ip block and we'll get into that so let's first start with what is subnetting subnetting is basically the process of taking a big ip network and we're going to work on that and i'm going to show you exactly what where how and why and breaking it into smaller components so think of subnetting as pre-virtualization virtualization i take a server with 128 cores and i chop it into 64 two-core servers it's just like subnetting subnetting i can take a block that gives me 16 billion or 16 million addresses and chop it into a whole bunch of blocks that give me 250 addresses so think of subnetting as basically virtualizing or chopping a big subnet or a big network into littler ones now why do we do server virtualization to conserve servers and maximally utilize them to get the best utilization why do we subnet to conserve ip addresses and use them maximally so here's a couple of secrets and i'm going to show you why when you're dealing with a server or you're dealing with a router or you're dealing with a firewall every interface has to be on a different subnet so if you've got a server with four network cards they all need to be on a different network or a different subnet so that is what we're dealing with so let's look at this architecturally let's talk about if for example in this particular environment i'm going to show you right now over here we have a router it's got four subnets on it now each subnet because it's got four interfaces like each arm on a router is an interface so what will happen is let's pretend i'm the router in the center and let's pretend my left arm over here sees the three subnet my white right arm over here sees the one subnet if traffic comes into me the router desks into the one subnet the router knows to take it from where i'm at send it through my body and send it out my right hand it'll reach the right subnet and then the traffic would go back the same way so that's the way the routers know what to do that's the way the routers know how to send these things so now let's pretend for example every interface on the router needs to be on a different network and the answer is we do so now let's look if we used a slash eight which is a class a address and we'll talk about these addresses each class i address basically subnet basically gets 16 million addresses so for example if we did this if we put 16 million usable addresses on one port on the router 16 million usable addresses on another port on the router 16 million usable addresses on another port on the router and 16 million addresses on another port of the router i would have used 64 million addresses for a single router i got to tell you you wonder how many ip addresses we have we have 2 to the 32nd minus 2. so wasting 64 million addresses for a single router would be kind of horrible so now let's talk about why we subnet so now instead instead of wasting 64 million addresses what if i could create a subnet that only has four usable addresses well actually total four addresses of which two are usable and all i needed to do was a link from point a to point b so i need one address over here and one address over here i don't need any other addresses for the subnet so here i've submitted my big my big networks down and i've chosen to use 16 addresses total for my four interfaces on the router now granted i've lost some addresses because we'll talk about it later the subnet address the broadcast address and other things but in this particular picture before my router is using up 64 million ip addresses in this situation with a little careful math and planning i'm using 16 addresses 16 versus 64 million now look at the number of ip addresses we have 2 to the 32nd minus two it's not that many we ran out of ip addresses a long long long time ago so because of this that's why we submit we have no choice subnetting enables us to take a big block or a big network and chop it down into little ones now what we're going to do is we're going to work through a lot of examples of how we do this and i'm going to give you lots of projects and then after we do the subnets and we work on the subnet design things that we'll do after that for which case we're going to have a little bit of fun are going to be as follows after we do the subnets then we'll design the architecture show you how the architecture and show you how all these pieces and parts actually are going to fit together in a way that's going to make sense for you but we've got to start somewhere and i just wanted to start by showing you those things so you understand why we subnet so now that you know why we subnet let's uh let's walk through the first part of something that's part of our history and if you don't know this history you won't be able to understand classless interdomain routing and we're going to have to talk about this because i want to make sure you know it so let's begin with this let's talk about ip address classes so when i started my networking career about 25 years ago we were getting out of something called classful networking and now we're completely out of something called classful networking but what is this concept of classical networking if you actually go into the ip addressing history you'll see that we had five classes of ip addresses and we had the class a addresses the class b addresses the class c addresses the class d addresses and the class e addresses and for the most part other than class d addresses which are used for ip multicast we don't really use any classes at all but what is an ip address class well anything where the first octet meaning that the extreme to the left was between 1 and 126 and the subnet mask was a slash 8 that was called a class a address and if you're dealing with real networking people like me that have been working in they all you'll often find people just talk refer to a class based upon the subnet mask which is not actually accurate so you might hear me call any slash eight a class a but it's really only uh if the first octet is one to 126. now a class b address is going to be that 128 to 191 and it's going to be a slash 16 which is basically 16 bits and we'll show you exactly what it's going to work out mathematically to network bits and ultimately 16 bits of host bits and then we had the concept of the class c address and these are your slash 24s but they were between the 192 to all the way to 223 and then we had our class d addresses which are used exclusively for multicast and that's 224.0.0.1 all the way through 239.255.255.255. that's your class d address and then realistically speaking your class e addresses were experimental they've never been used for anything so now you know the type of address classes we're talking about and now that you know the address classes that we're talking about let's kind of start walking and looking through so if we actually go back to this architecturally and apologies for hiding behind so many slides today i'd like to be right out there front and center with you but you know today there's just some things we've just got to visually present so apologies for all these extra slides typically speaking this is a class a the top over here and the reason a class it's a class a is a class a has eight bits of network and 24 bits of host so when you get that 10.0.0.0.0 8. that's you're getting those.0.0.0 those three things that's your 24 bits of host mask that you can actually use for as opposed to the network bits which are just the first part so realistically speaking that's what we're kind of talking about our class b addresses are going to be balanced they're going to be 16 bits of host and they're going to be 16 bits of network and our class c addresses are going to have 24 bits of network and then 8 bits of hosts and by doing that ultimately what you'll see is we're going to run into an issue of addresses so 24 bits of host and eight bits of network as we're dealing with class a when we're dealing with class b's we're dealing with eight bits i'm sorry 16 bits of networking and 16 bits of hosts and we'll just modify that um not really sure how the graphics got this way but uh apologies um and then ultimately we'll have 24 bits and network masks and eight bits of host addresses that we're dealing with in this section so now you can see what's kind of going on now that was classful routing now the concept of classless interdomain routing is a little different we're basically saying we don't have to stick with these bits these pre-programmed bits we can chop this down any way we want and that's really what it is so instead of a class a being eight network bits and 24 host bits what if we could move the network bits over maybe we have 24 network bits and eight host bits or maybe we have 16 bits in network host bits what's going to determine that well how many hosts do we need per subnet if i've got a point-to-point link meaning a wire on both sides and each interface has to be on a different subnet i can never use more than two ip addresses on both sides of a wan link never more than that so my subnet should never be more than two usable addresses for that so that's going to be a slash 30. but i might have another environment where i need 230 servers so that's going to be a slash 24. so what are we going to do we're going to figure out how many hosts we need per subnet and that's going to the way we're going to determine how we chop it so you guys are all going to get them look good in the math if i need two hosts i pick one subnet site if i need 200 hosts i pick a different subnet size if i need 400 hosts i pick a different subnet side want to know the secret to subnetting do you want to do the two subnets you probably remember in your head and if you knew none others you'd probably be pretty okay we use a slash 30 meaning two five five two five five for point-to-point wan links you know what we use for land links almost exclusively 24 which is a 255.255.255.0 i will tell you later architecturally why those are the two that are chosen i will show you why we do this how we do this and why it's the right choice for an organization and 99 of the cases without even thinking about it but i just want you to know that we're going to get into the math we're going to get into the fun of course we'll map it out architecturally and then because it's networking and networking can be fun if you want to play with some bgp or ospf all these things along the way in the end we'll do it so let me know that you're still here type hashtag cloud hired in the chat box i know you're here i know you're all awake alert and oriented and then we'll get on to some more fun so please hit the like button tell others to join it tell others to subscribe you know we work pretty hard to uh try and make sure we bring as much content as we can for you especially free so it's always good when the algorithms all see that people are happy and having a good time so excellent and thank you so now i want to go into this concept of exactly what determines what we choose when we choose it and how to choose it it is the number of hosts that you're going to need for the most part that drives your subnet size the number of hosts that you actually need so make sure you've got the right number of hosts in your software and you also need to know the number of subnets that you're going to need and that's going to give your mathematical ability do you chop on one side or you chop on the other side how many hosts you need now there is a formula to determine how many hosts and i'm going to show you the good mathematical formula formula and then i'm going to show you the i'm bad at math cheat formula but in any case i just need you to know how to do it i need you to know how to do this how to understand how it works look once you get experience you can build yourself an excel spreadsheet or use any tool you want but you have to know how to do this because it's super uncritical so let's go to this environment how do you know how many hosts are going to be available well we're dealing with binary numbers so basically binary on or off zero or one so because we're dealing with a binary address it's either going to be a zero or a one zero off one on realistically speaking that's it so here's the cheat and i'll show you in a minute if you want to know how many hosts per subnet the best thing for you to do is take the number of host bits to the exponent of of host bits and i'll show you what i mean by that and subtract 2. but i'll show you another cheat way too so if for example i'm going to make a 30 i've got six subnet bits right eight host bits because we borrowed some you know six seven up that's two host bits so in a slash thirty how many will have two hosts so it's two times the number of host bits so two times two which is four minus two okay wait now instead of a slash 30 let's say i chose to use a slash 29 okay so now that's five subnet bits and now three host bits so in this case if i want to figure out how many hosts would be two to the third because i've got three host bits two exponent number of host bits gives me eight minus two equals six okay now instead of doing the slash 29 i want to do a slash 28. how many hosts do i have 2 to the fourth why 2 to the 4th i've got 4 host bits and 4 network bets so 2 to the 4th 2 to the 4 4 toast bets gives me 16 minus 2 gives me 14 usable addresses now there's something goofy about aws and some of the cloud providers will talk about that later but i want you to get the subnet mask first instead of the slash 28 we choose to use the slash 27 slash 27 oh wait that's five net that's how many how many host bits five sets of host bits and three sets of network bits so now two to the fifth is 32 minus two and that's what we're dealing with here's how we're actually doing it we're dealing with two to the number of hosts minus two now when we're dealing with aws we're gonna have to deal with some other funny business aws decides to reserve the first four addresses of a subnet as well as the last so realistically speaking with aws it would be two to the x minus five as opposed to two to the x minus two and when you're dealing with the cloud providers they all have something goofy with their addressing schemes just please remember theirs i'm teaching you the regular networking thing that's going to work on your cisco router your juniper router your palo alto device your fortinet device or everywhere in between and what's going to be going on in an organization's data center and you're going to have to deal and contend with both so you need to know both and that's why we're spending so much time and effort on this because it really matters so that's why we're dealing with these things so let's walk through the math now we'll get into the nuts and bolts of how this works so let's say here we have a class b and what does this look like naturally over here let's say this was a 172.16.0.0 16. this is what it started out us this is where we begin so you know that's where we begin now then that's going to get subnetted down and how do we subnet it down as it stands right right here we've got 16 a class b is going to have 16 subnet 16 network bits and 16 host bits so if you want to know how many hosts would be in a subnet for class b it'll be 2 to the 16th minus 2 which is a pretty big number it's like 65 000. now if we don't need 65 000 hosts in a subnet and guess what everybody you can't put 65 000 hosts in a subnet in fact you don't even put a thousand hosts in a subnet and the reason you can't put that many hosts in a subnet is as follows all your computers do this i'm here i'm here i'm here arp arp our broadcast i have this service i have this service they broadcast computers they chat they get talk and the subnet and they fill up the subnet with traffic the servers and the systems and the systems have to process the broadcast traffic it sends the cpus to the ceiling so you can't put that many servers or computers in a subnet anyway because these computers especially windows devices and apple devices they send all these broadcasts and they constantly fill up the system so the point is you can't put that many hosts in a subnet no matter what you'd want to do because the host themselves will actually fill up too much it's like putting so many people in a room if you had 100 users in a room it might be okay if the room's big enough but if you put a million users in a room you're not going to hear anything because the people are going to be talking so loud if you wanted to have a phone call a polite conversation with someone next to you you wouldn't be able to so with your systems you're never going to put more than 500 systems in a subnet and the reality is you're almost never in your life going to put more than 250 units in a subnet so your normal lan subnet mask is going to be a slash 24 and your normal lan subnet mask is going to be a slash 30 but there are times when you need to do more and we're going to have to route aggregate up and chop down for routing purposes so you need to know how to go in all different directions so now 16 bits is subnet of network mask 16 bits of host mask but let's say we choose to borrow six eight bits of host mess and give it to subnet masks so now what do we have let's go back here where was i bear with me a second so let's go back to this situation what this looks like if we don't subnet this down all these are going to be zeros this is what we start out with this is what it's going to look like initially that's our class b address and we decide to subnet it down and we make these subnet disks so now let's say we've got these subnet bits and we change them all to one let's make sure i've got a to the 1 2 3 4 5 6 7 8. so now what does this look like to you this is a slash 24 why is it a slash 24 it was a slash 16 so we have 16 bits of host of network bits now we have 16 bits of subnet bits which get counted towards the network side so they get shifted towards the networking so now we have the 172 168.0.0 24. we submitted it down how many hosts can we put in the subnet now that it's a slash 24 8 bits of host 2 to the 8th minus 2. come on everybody do it 2 to the 8th exponent eight on your calculator the number you get is 256 minus two how many does that give you it gives you 254 addresses one to know the cheat way to figure it out if for example you know your mask so let's say for example you take 256 256 bits so let's say for example you take 250 you take 200 let's say you take 256 and let's say we go back to that slash 30 which is minus four 250 and you know that's going to make it more complicated for you i have a cheat approach to math and then the cheat approach worked well for lots of people um but i maybe let's just keep it focused to the main approach but you'll be if you take 256 and you subtract your a bit your basically your hope is you'll ultimately get to the same destination but let's keep it simple and let's keep me all in one formula the standard formula that it's going to work everywhere instead of the old network engineers formula so just let's keep submitting it down so let's let's extrapolate this into another environment i want you guys to all understand this clas slash 24 and if it's at 172 16 or 172 168 doesn't matter it it's still a slash 24. um but in this particular environment let's say we've got two five five two five five two five five zero this by the way is traditional subnet mask now what you see over here is something called cider notation so what is cider notation cider notation is for people like me that have been working with classless inner domain routing for a while and we just think of terms of bits so realistically speaking you could look at this subnet mask as 255 255 255.0 or you could look at it as 24 bits of network mask it doesn't matter now what this is which is the same as this mathematically is all ones for the first octet all ones for the second octet and all ones for the third octet now that's what we see this is physically what's going on mask wise bit wise but this is what it's actually going to look like when you would type it in but this is actually what it would be referencing in terms of bits of subnet mask look 8 plus 8 plus 8 equals 24. now let's say we wanted a shrink of the slash 24 down to a slash 25. here's what we would get it'd be the same ones ones and ones but here we're going to go one direction over hey arun's so good to see you caught architect and so again that's going to be popping in so now let's say we want to add two bits of subnet mask because we want to slash 26 and as you notice as we shrink down this way our subnets are going to get a lot smaller they're going to get much smaller so now we're going to add two bits of additional submassive masks over the traditional class c so now it's going to be two five five two five five one nine two and i'll show you where those numbers come from and it's a slash 26 etc etc now here's an extreme one something that only network architects like me will ever see and this is called the slash 32 subnet mask this is a host mask just a host route so for example those of us that have putting loopback addresses on routers lots of us use the slash 32. why waste no ip addresses but this is something that nobody will ever use you can't use it outside of a loopback address on a router so kind of get that out of your mind so okay are we all ready to get started let's do this let's work through this all together so when we're dealing with subnetting what are we going to be dealing with we are dealing with the network bits we're dealing with the host bits now with every network we're going to have something we're going to have the addresses that need to be tossed away so let's talk about the addresses that need to be tossed away there are in normal networking the first address of a subnet is used by the router or as the network number and the last address of the subnet is used for broadcast meaning send something that everybody else can see so i need to make this clear so let's over for argument's sake let's say we take a slash 30 subnet let's say we've got a 192 168.1.0 30. this is the subnet what the routers will use to put in the routing tables they will have a 192 168 1.0 30. this will be the route and this is the that's going to be the subnet then we will have the 192.168.1.1 slash 30. that will be used by that's one usable address we will have the 192.168.1.2 slash 30 that's a usual address and we're also going to have this 192 168 1.3 address now this dot 3 30 is a special address it's called a broadcast address and what makes this address so special is is this is what the routers keep in the routing table this and that's how they know where the subnet is so this can't be assigned to a host because it's used by the router this is assigned to a host meaning one side of the link this is assigned to another side of the link now this here this all one's broadcast that says something that can be seen to everybody so if i send them a ping to 192.168.1.1 192.168.1.1 is going to respond with an icmp echo reply if i send a ping to 192.168.1.2 192. 168 1.2 will reply to me and say got it mike now if i send a ping to 192.168.1.3 here's what's going to happen the 192.168.1.1 and the 192.168.1.2 will both respond to the ping because i sent a broadcast which is hey everybody answer me and everybody says i hear you mike i'm answering you mike so that's what's going on so that's where we look at these things so the first address of all subnets is used by the router the last address is used by the broadcast now if you're with aws aws actually reserves the first four ip addresses and then the last one which means could you use a slash 24 i mean a slash 30 in aws except for the link on the way to get there well slash 30 is for addresses aws reserves five you can't use the slash 24. i mean it's 30. in fact the slash 29 would also be problematic because that would give you what it would give you 2 to the third which is 8 minus 5 is three again you'd be in trouble so the smallest subnet you can use with aws is a slash 28. now here's the thing best thing you could possibly do lan segments slash 24 when segment slash 30 you don't have to think about it as much that way generally good things if you can do that but you know keep these things in mind so now we have to i want you to all with me we're going to do some subnet math we're going to look at these random addresses and we're going to figure out where does the subnet begin where does the subnet end what are the usable address ranges and what's the broadcast address so here's how we do this so if any of you guys get some ccna math problems see this on an aws exam or a google exam no it gets pretty easy let's do it so let's do this 192 168 1.132 okay somebody show me what those 25 bits of subnet masks looks like in the chat box and and so put it in octets you know 1.1.1.1.1.111111 etcetera dot 111 dot etc etc somebody show me what does 25 bits of subnet mask look like in both cider notation as well as decimal notation i'll tell you what i'll kick it off in decimal notation and somebody else show me in binary notation decimal notation it's going to look like 2 5 5 two five five two five five dot one two eight that's what a slash 28 is going to look like everybody so i'll pop it in there now i want all you guys to show me that mask in subnet bits or at least somebody the first person that does it you know gets the gold star for the day show me what that looks like in binary notation david page excellent job perfect you showed it to me exactly in binary notation um caroline excellent a jenny almost um only one bit of subnet mask in the last not two in the last octet a seam great job okay this is good stuff arun excellent i'm so happy to see you here arun abraham excellent jazz well we're gonna have eight in each box but yeah i figure that's your short hand but yes anson excellent excellent i'm super happy to see that okay so now let's have some fun with this and broderick it'll be the same but we're gonna have one one one one one one one one dot because it'll be eight each time great for you old time honey official we're doing great i'm really excited i love seeing this i love people us watching people do a really great job for the day fantastic everyone so now let's go back to play with our math so now in this particular example i was giving everybody let's go back and let's have some fun with it so now let's figure out where this subnet begins here's the secret make yourself this chart until you get used to it 2 to the eighth is 128 2 to the seventh is 64. to the sixth is 32 2 to the four fourth that's 16 2 to the third is eight two to the second is four to the first is two and we'll just ignore this whole one thing over there so let's think of it this way so now 192 168 1.132 where does that fit well let's now go to our cheat sheet does it fit inside of 128 is one is 132 divisible by 128 in some way shape or form with a number that is greater than one yes it is so does 128 fit into 132 the answer is yes so the subnet for this first one is going to be can anybody tell me where the subnet begins i'll do it with you 192.168.1.128 that is the subnet oh it's a slash 25. that's where the subnet begins where does this subnet end everybody the subnet's going to end at 192.168.1.255. what are the usable addresses everything from 192.168.1.120 all the way to 192.1 0.168.1.254. so these are kind of you know one of the things that i want us to do that i want you to understand so i have a couple thoughts we're going to do a math problem or two here i'll open it up for some questions we'll do some more math problems we'll open up for some more questions because i don't want anybody to be lost so if you guys have any questions start asking them what will happen is chris from my team is going to aggregate them and say every 15 or 20 minutes from now on we're going to come off of you and answer your questions because we're going to build on this stuff relatively fast once i know that everybody gets it and we're going to get into some architecture and some things that could actually get potentially advanced today if you guys catch up on the math so let's make sure we answer any questions so now let's do another math problem for this slash 27 let's say 172 16 1. 120 27. so what is uh 27 bits of subnet mask look like well for one thing it's gonna be two five five two five five two five five plus this plus this plus this that's going to be your subnet mask so what does that look like 128 plus 64 everybody what's up 196 add 32 what does that get you to two to four so there's that so let's go back to this so now let's say we're we're here so okay so now working on this assumption where does this subnet mask go where does this subnet begin 170 216 1.128 we've got this is the mask and the same hands excellent you've got it jazz excellent you've got it that's what the mask looks like excellent excellent excellent stuff super happy to see this so now that i see this here here or here now what do we do and christina lee i'll show you how the people calculate it so quickly um in a second they don't have to do too much thinking about it so christina the reason it's so fast is we know we've got 24 bits so that means the first a bits are all ones it means the second eight bits are all ones it means the third eight bits are all ones and then we've got three more bits and we go from right to left yeah so i'm sorry from left to right when we calculate them so the point is is 24 bits will always be one one will be eight ones dot eight ones dot eight ones dot zeros a slash twenty five will always be eight ones dot eight ones dot eight ones dot one one and the rest zeros uh slash 30 will be for example always eight ones eight ones eight ones dot six ones with two zeros afterwards so realistically speaking that's what's going on as people are just looking at it saying okay i just remember one bit two bit three bit four bit five bit six bit seven bit and when you just remember that way it just sort of clicks in your head so let's calculate this one if we know the mask is gonna is gonna end in a two two four and how many addresses do we have in a slash in a two to four well we can take 256 minus two to four and how many addresses that give you for one thing 256 minus two to four that gives you 32. 32 minus 2 equals 30. or do you wonder how many addresses you could do you could do to the 2 to the 5th minus 2 it'll still give you the same addresses so now let's look at the subnet mask everybody where does the subnet mask begin 172.16.1.27 where does this thing start okay is it divisible inside of 32 the answer is no but that's how many 30 that's how many bits we have so we know this subnet begins with 172.16.1.0 27. that's the subnet that's the beginning so now let's figure out where it ends what is two to the fifth two exponent fifth everybody two four eight sixteen thirty-two so that is your broadcast address 172.16.1.31 why is it 31 and not 32 because we start counting at zero zero 231 equals 32. 1 to 31 equals 31 0 to 31 puts us in these kind of positions so that's realistically speaking that's how we calculate these things so now this means that our subnets the usable addresses in our subnet are going to be 172.16.1.1 all the way to 172.16.1.30 and then this will be used by the router this is what's going to be put in the routing table try and get this this will be in the routing table this will be the broadcast address and these will be the address that are usable for your hosts let's do one more now let's do this let's go to let's go to this 192 168 1.17 28. okay everybody what's a slash 28 look like let's go work out those bits it's going to be eight bits plus eight bits plus eight bits plus what four more bits slash 28 eight plus eight plus 8 is 8 times 3 is 24 plus 4 more that gives us 28. so tell me here where is this subnet subnet going to be going to begin 28 is 17 divisible by 16 more than once and pierre great job i'm loving it you're correct a seem excellent ronald awesome cloud fresher caroline excellent ansen tom excellent alex wood excellent for the last octet you got it you've got that last octet exactly christina lee excellent femi power okay this is making me really happy really pal david page you guys got it you guys are doing amazing okay i'm just so happy this is going to make my day i'll be in tears by the end of the day if you guys all get subnetting and super knitting you guys get this right it changes everything so i'm so happy for you guys okay so now let's do this let's go back to this address somebody tell me where the subnet begins and then where it ends so we now know from the math that you guys all did that six that 17 that 17 is divisible by 16 by at least one so that means this subnet is going to begin at 16 in this particular case so let's go up here and now let's start with the first ip address it's going to be 192.168 at 16 is the network mask meaning it's used by the router this is the network which is used by the router now somebody add 15 more to this 16 and then we know what the broadcast address is going to be it's going to be 192.168.1.31 this is the broadcast so now what addresses are usable somebody tell me what addresses are usable type it in there in between the dot 16 and the dot 31 what are those addresses somebody tell me in the box what addresses are usable someone gave me a range like a dot something to adopt something else pierre i hope i clicked old time official you still don't get it that's okay we're going to do a few more of them for you evo 17 to 30 a seem 17 to 30. now amrat i'm real happy that you're coming off a mutant you're putting that in there excellent job amrat very happy okay so good so let's ask this question again so abraham has this question it's a really good question and i know we talked about it but let's talk about it again how do we determine the number of hosts per subnet well very simply we do the following we determine the number of host bits and then from there we just do two to the exponent of the number of host bits minus two and then we get to our answer so let's say we have a slash 24. if we have a slash 24 how many host bits do we have we have 2 exponent 8 in your scientific calculator which gives you 256 minus 2 gives you 254. now let's say instead of a slash 24 for which case we've got eight host bits we've got a slash 25. how do we determine how many hosts we have in a slash 25 we do a two to the seventh because we've got seven host bits two to the seven equals 128 minus 2 that means you've got 126 hours per subnet so now the next thing you might want to know what would your next subnet look like let's say for example instead of the slash 25 we just talked about we have a slash 26. that means you're going to have the slash 26 is 26 network bits and 6 host bits so we do 2 to the 6 which is 2 times exponent 6 on a scientific calculator two to the sixth gives us what 30 or 64. i should i think two to the six is so two to the seventh is 128 two to the sixth is 64. so we do two to the sixth which gives us 64 minus two now let's say we want to shrink it down even somewhere so instead of that slash 25 or the slash 26 we want to use a slash 27. so now we've got five host bids so that's going to give us 2 to the 5th meaning 2 4 8 16 32 minus 2 or 2 exponent 8 in your scientific calculator so that's pretty much it that's how you're going to determine how many host bits so the next question is from pierre and it's a great question how do you determine the starting address now this is something a lot of us put on interviews because this gives us the answer if somebody understands this so let's work through it um because it's a great concept so pr this one's going to be all on you and i'm going to have you respond to this in your own time and i know there's a delay so take your time so here's what we're really talking about and anonymous yt we can't just use a calculator and see here's the thing what happens when people just use the calculator is they're having something else that does the math for them but the problem with that anonymous yt is the people that are using the calculators for the math if they don't understand why the subnets are being designed the way they are and why they're being laid out the other way then they're just using math that works and if we don't know to design the ip addressing scheme and where the ip addresses need to be in the network it will negatively impact the routing and i gotta tell you in about 25 years of networking we'll say about 80 percent of the problems that i've seen are from people that actually use subnet calculators because they actually didn't understand the networking design they were going to do put the wrong ip addresses there which messed up the ospf the bgp the intermediate systems intermediate systems and then everything basically piled onto the bad foundation so i'm not going to say i like math i hate math and anonymous yt and you know i'm not going to say i haven't built myself an excel spreadsheet that i would use for an ip addressing plan that i use for all my systems but i build it based upon the network and the systems based upon the designs so you realistically need to know how to do this plus you know the ability to work without this is going to be really challenging so i want you to be really strong in this because i want you to have a good career without it it's going to be near impossible so let's try and deal with this so pr let's go back to the octet piece so if it's a slash 8 we already know what it looks like if it's a slash 15 16 we know what it looks like so let's just do these masks one more time so let's work through this so let's say for example we decided to use let's pick this one we're going to use a slash 28. so first let's map it out here one two three four five six seven eight that's eight bits assuming i did my math correctly one two three four five six seven eight one two three four five six seven eight so eight times three gets us 24. now let's add four more bits here and then the rest are going to be zeros so now we know over here um exactly why what this is going to look like this is the subnet mask now pierre if you want to know where it starts um and uh jenny i'm a ccie my cci number 7417 cisco taught me to use the slash 30 um for point-to-point links um and you can't use a slash 31 for a point-to-point link because you don't you don't have anything for the network and you don't have anything for the broadcast so however they got misinterpreted i don't know but uh cisco only says slash 30s and when i was in cisco that's how they taught us to do it and when i worked and trained systems engineers at cisco um on cisco systems engineer tv we always use slash 30. so i don't know about the slash 31 because it just won't work so unless they added a new feature that would enable it to break routing or traditional routing and which i've not configured the routers for a couple years but let's go over here so we've got our 8 bits our 8 bits our 8 bits and our 4 bits so now let's look at this so where are our eight bits coming from pr so we've got a 128 plus a 64 plus a 32 plus a 16. that's what our mask is going to look like now let's go back to this slash 28. what you really want to do is you want to see if your things are divisible by the mask so for example if you are here is 19 divisible by 16 and the answer is yes if 19 was if if 19 was not divisible by 16 let's say for example in this environment it was o8 if the if the subnet is divisible by 16 and it's not greater than 16 we know we're starting with the first subnet or the zero subnet the beginning part of it and then we're going to be working our way all the way up to 15. why is that because this number is the number of bits we have and the number we had is less than that so it's going to start at 0 and work its way up to 15 which is the broadcast now pierre let's pretend this was 40 49 is 49 divisible by 16 pr yes it is it's divisible by three times plus the number one so if that was the case pierre the subnet would begin it's at a 48. now it would begin at 48 and it would go up 16 more so that would be 49 50 51 52 53 54 55 56 57 58 59 60 61 62 and 63 because remember we start counting at zero so that my 15 plus the zero which equals my 16. now by comparison pr let's say it began with 78 is 78 divisible by 16. sure it is 4 times 64 is 4 times 16 is 64. 5 times 16 is 80. so we know we overshot it with 80 pr so let's go back to 64. we can look at this subnet and say this subnet begins with 172 17 1.64 and add 15 plus that which means we know where the subnet's going to end it's going to be 70 it's going to be 60 floor plus 15 and that'll be where your sun that ends so that's really how we're doing that pierre did that make sense to you if not and if if other people need that let's go back and do it i want to make sure everybody gets this could you guys help me out and let me know if you got it if you got it type cloud hired if you want me to cover it again say cover it again in the chat box just so i know did you want me to bring in any other questions um if we got questions yes um although i it looks like some people want me to cover this one again um so let's make sure we cover this one more time okay can i go over the host bits if we go over more than one octa okay so let's let's now let's go back and look at this one more time our bits are going to be counted in octets with ip addresses we have a 32-bit address which is covered by eight bits and four octanes so a slash eight mask looks like this so that's a slash eight a slash 16 looks like this i'm not counting my zeros so i could potentially be typing too quickly or missing one and a slash 24 looks like this any time we add networking bits we go in this direction so if we want to go to a slash 25 it's going to be this if we want to go to a slash 26 all we're really doing is we're just moving over one more and we're moving from left to right but that's what's really going on so now that we remove that from the equation let's figure out if you want to know where your subnets start first you need to know where your masks you need to understand the subnet mask and then you need to understand how many hosts per subnet so if we need to draw it back let's go to this one this 172 17 3.129 let's first draw the mask so you can see what it is one two three four five six seven eight one two three four five six seven eight one two three four five six seven eight that gives me 24 right now i've got 29 so 24 needs how many more it needs five more one two three four five one two three so now let's do this let's count over one two three four five so now we know exactly what we've got in this slash 29 and this slash 29 obviously won't work with aws but it would work in the rest of the networking world so with these five host bits where does this network begin well i don't know we're going to have to calculate it out so we have three bits of host bets what is two to the third everybody two four eight what's two to the third everybody eight now let's go back to nineteen is nineteen in some way shape or form greater than eight is it divisible by eight yes if we divide 19 divided by eight and we the reason we know it's eight because it's two to the third because that's the number of host bits we have two four eight so now 19 is divisible by eight how many times two nine times two or so so so 19 is divisible by eight how many times eight sixteen twenty four so greater than 2 but less than 3. so now let's think about this if the subnet begins if it's greater than 2 and less than 3 that's our subnet range which means now let's think about this where does it start where does this 8 times 2 begin so we've got a 172.17 dot 3 dot what is that weight if we've got multiples of 8 and it's greater than 2 and less than 3 what is 8 times 2 everybody 8 times 2 because it's divisible by 2 as a whole number is going to give us 16 nitro pen that's exactly where the subnet begins now the next subnet is going to begin the next network is going to begin eight numbers higher so the next subnet is going to begin at 24. and then the subnet after that will exist at a 32 and then the subnet after that will exist at a 40 and then the next one will be a 48. so now let's going back here where is and nobody messes just because we all mess up a number it's no big deal mark i mess up lots of numbers so now if it begins at 16 and then we know the next subnet begins at 24. where does this thing end it's going to be the largest number that's not divisible by 3. so 24 is divisible by 3. what's that number that's 1 less than 24 well alex you're doing pretty good 24-1 equals what alex 23 exactly so the subnet will end 172 dot 17.3 good job alonzo um fenney power alex wood i can make a number off by one two or even a million and not even know it so you know you guys are doing great i mean this is networking it takes everybody a little bit of time to do this you guys are doing fantastic um we're gonna do this we're gonna make sure um that you guys get it in broadcast you're right the dot 23 is the broadcast exactly exactly exactly exactly so good so let's keep doing this also i'm being told that if you guys are having a good time if you can hit that like button subscribe i'm also told i should remind you guys that on monday we're going to have a free how to get your first cloud architect job webinar and it's going to be a really really great time we're going to teach you everything you need to do to get hired but i'm going to go back to the teaching i'm learning how to use youtube so if you can hit the like button subscribe hit the bell all these metrics that i'm supposed to know all about is somebody that spent some time on youtube but seriously it does help our youtube metrics which helps us teach more people around the world and we do a lot of free training and it means a lot to us so anytime you guys can share our message of free training to help others in the world it really means a lot to us so let's go back to uh here now let's go back to this situation the dot 30. okay somebody show me so what's this dot 30 look like well let's go back and redraw our mask we know the first three octets are all ones right and then we know we count from left to right so dot 30 means we've got 24 so that's 25 26 27 28 29 and 30. let's add two host bits so we get the 32. so now and thank you so much nitro pam we love having you here as well and i love sharing knowledge and experiences so now let's go figure this out we've got six bits of subnet mask everybody so one bit two bits three bits four bits five bits six bits okay here we go tom excellent you got it so now going back here is two divisible by 4 with a number of greater than 1 no 2 is not divisible by 4. i mean it would be is 0.5 but it's not a whole number so this subnet is going to begin tell me somebody where does the subnet begin given that 10 given that 2 is not divisible by a sum number so we're going to begin at the first subnet and we're going to end at the last part of the first subnet so somebody tell me where the subnet 10.0.0 dot something slash 30 begins well tell me where the 10.0.0.2 30 subnets somebody tell me the first address tell me the last address evo this looks like something great for you because i think you're you're miss math but orthius just got it zero is the network three is the broadcast um one and two are the usable addresses so exactly exactly exactly perfect tom do you speak greek by chance with your name if you do just say yasas in the chat box and i'll know that you speak greek otherwise you know super wonderful having you here i'm just always looking for more people to speak greek to so okay let's uh let's go over here let's do another one so now we've got the 10.0 24 everybody slash 24. now on our slash 24 for the fun of it how many hosts okay well how many bits is it going to be two to the eighth two four eight sixteen thirty two sixty four one twenty eight two fifty six here we go there you go so now you know that's uh the number of posts two fifty six minus two because remember gotta lose the broadcast gotta lose the subnet you wanna know how many hosts you have in aws 250 2 to the eighth minus five same thing you just have to learn which addresses you're going so going back to this one going back to this 10.0.0.9 we're having our fun we're doing our things we're addressing our stuff all kinds of cool let's go look now where does this subnet begin it's a slash 24 which means 254 usable addresses but a total of 256 for the subnet so is 9 divisible by a whole number into 256 or 254 no since 9 is not divisible by 254 for example where does it begin the subnet's going to begin well nitropen good job exactly the subnet will begin with 10.0.0.0 and it will continue all the way to the broadcast address which is 10.0.0.254 which means the usable addresses will be the dot one all the way to the dot 253 and the range that you guys have listed of zero to 255 is perfect because 10.0.0.1 is usable by the hosts and 10.254 is usable by the host the dot zero like amaranth said is used by the router and the dot 255 is used for the subnet directed broadcast so you guys did it perfect perfect perfect perfect perfect now we're going to give you guys one more and then we're going to start doing going in the opposite direction why are we going to go the opposite direction because opposite direction is routing this direction is addressing the opposite side is routing and when we get this and then we get the routing we can speak architecture and then we can map out some network architectures which you'll see commonly in the network environments the cloud environments after all cloud's nothing more than a network and a data center that's been virtualized it's just somebody else's computer and somebody else's network when you take all the mysticism out of them now let's make sure let's make this a little bit harder let's take this one and let's make it 14 okay so now in this particular 10.1 to 30. i made this one an ugly one because we're going to challenge your math skills and if you're like me and you're mr verbal and you don't like math well then let's just say you're going to think about this one but you all got it and it's not that complicated so it is a slash 30 which means 30 bits of subnet mask so 25 26 27 28 29 30 okay so that gives you the number four two to the exponent of two gives you four so now where does this begin is 46 divisible by four yes how many times 12 times 4 is 48 11 times 4 is 44. okay so where does this sum that begin tom ortheus polioreo it begins with a dot 44. that's the subnet it ends with a dot 47 which is a broadcast and dot 45 and dot 46 are the usable addresses so you guys got it you guys are doing great i'm so happy give yourselves a round of applause pat yourselves on the bat you guys are doing really great so now you know this is good stuff everybody super happy for you so now we've done some really nice networking stuff we've subnetted it down now we're going to have to go the other direction why the other direction why is the other direction so cool because now we're going to start talking about routing okay so now that's thinking about networking what do routers do routers build the map they determined to go here because to get to john cells go out this interface to go to chris's house go out this interface to visit my cat cindy go over here but see routers don't think in terms of my cubecut cindy and they don't talk about you know how to get to chris's calf they think in terms of subnets so realistically speaking the routers are just building a map of subnets and now you know you and i in humans we think of main street and villages and things that make sense to us the computers just look at these subnets so now realistically speaking what we're going to actually have to do is we're we're going to go start and we're going to supernat them and for the question from a scene hands before we actually do this and actually before we do this let's make sure we address all the questions so let's address the questions before we go into the next section so the question from naseem huns how does the lesser number of usable addresses given play in the last example it really doesn't change much of anything a seam so what it means is because of the way aws uses addresses you're not going to be able to use a subnet that's going to be smaller than a slash 28 but the math is going to be the same the only difference is normally speaking in networking if we go back to any of these examples let's say we go back to this example let's go let's go back to this example over here because it's a valid question let's say we go to this um address over here the slash 28 the first four bits are always going to be eight the second four bits are always going to be eight these four bits are going to be eight and then we're going to have four bits of subnet mask here and four bits of host mask so what is the subnet gonna be on this a seam it's gonna be the same one where does the subnet begin and what is gonna be used by the router well let's go to this environment let's say we go to this 28 that we just picked so 25 26 27 28. so now let's figure out what is the visible by 60 by by 16. is 78 divisible by 16 the answer is yes 4 times 16 is 64. 5 times 16 is 80. so that means that the subnet is going to begin at 4 times 16 which is 64. so we've got this 172 16 dot zero we've got this one so i'm sorry home 172.17.1.64. this is the subnet that's what the broadcast is add 15 172.1.16.1.79 that's the broadcast it's the same on either one so it seemed none of this changed now the only difference is aws won't let you use dot 65 dot 66 or dot 67. so your your your the addresses that you'll actually be able to utilize with aws will be less because you won't because this is going to be reserved reserved this is going to be reserved and you know the next one's going to be i'm sorry i'm trying to i've switched from a mac recently to windows and i'm still learning how to use windows you used either linux or mac for the last couple of decades um and all of a sudden i'm using windows because my mac couldn't handle my needs and let me tell you it's it's it's a change so okay so we're here so that's realistically speaking what's going on here is you still have the same subnet you still have the same broadcast you just can't use these three addresses so a seam it's not going to change anything you just can't use these so there's there now let's see remember what i told you slash 30s for wand links slash 24 is for wind links and when you're dealing with the slash 24 instead of having 254 you're just gonna have uh 251 so it's really not going to be anything that affects you so i hope i answered your question there before we move on were there other questions on this topic okay well sri welcome i love networking too now let's go the opposite direction everybody exactly there's going to be less addresses because they uh waste them so if there would be a total of two to the fourth um or if you would typically get two to the fourth minus two it'll be two to the fourth minus five same thing nitro pen you just get to use less than now what we previously did everybody was we borrowed network bits uh we we borrowed host bits to cre we took the network and we added some more network bits to effectively speaking increase the number of or to decrease the number of host bits to decrease the number of subnets as opposed to decreasing the number of hosts now let's go the opposite direction what if instead of taking a bigger network and making it smaller i could make a bunch of smaller networks look like one now this is kind of the this is the point where you're going to be like okay great mike just taught me how to chop big networks into small networks to save ip addresses and now it's going to teach me how to take these smaller networks and then build them back into a big network okay so this is pretty important for you to understand we maximize the use of ip address space by taking subnets and shrinking them down we optimize our routing by minimum minimizing the number of routes in the routing table and engineering our traffic where we use a specific interface for a specific subnet or another interface to go to a specific subnet now the ability to determine whose traffic goes where and engineer your traffic is going to be explicitly based upon how intelligent your addressing scheme is so remember we've just gone this way we've been moving in this direction we've been adding network bits and and by doing so we've been taking the host bits and creating network bits now we're going to go the opposite direction okay everybody how many hosts are available in this subnet 192 168.0.0 24. so if you don't know do 2 to the 8th minus 2 and tell me 254 exactly pierre exactly mark okay good so now that you guys are getting it i'm david page ashley hughes tom femi ashley excellent now since i know you guys got that now somebody tell me clayton um tell me how many hosts are going to be available in the next subnet up 192 168.1.0 24. again it's going to be 2 to the 8th minus 2. so how many guys have already done the math just do it again one more time for me just so i know you understand exactly it's a class c so it's going to be 254. so now this 192 168 2.0 is going to also be 24 and the 192 192.168.3.0 is also going to be a 254. you guys get it now let's look at this now if what we previously did so let's go back and let's add some cool math and we used to do this so right now we have three four five six seven eight one two three four five six seven eight dot one two three four five six seven eight dot one two three four five six seven eight okay now let's do this this is the one subnet mask for the top subnet next subnet next subnet next submit hey wait don't they all look the same so now can you guys all see this does everybody see that we've got four contiguous subnet masks this octet is zero this one is one this one is two and this one is three does everybody see that okay so now let's think about this if oops this windows computer it takes some getting used to it runs great gotta say um these threadripper cpus are pretty awesome so slash 24 i hope so number one never mind apologies here i i did some little funny cutting and pasting one one three four five six seven eight the third octet should be all ones thank you amrat this is the danger of life when you do live and you're doing your own math at the same time as you're typing you do miss stuff something tells me let's just make it over here so i've missed a number one two three four five six seven okay so you know it's looking relatively as clean as we're going to do for live so now slash 24 24 0 and 1. instead of this being a slash 24 and this being a slash 24. now what if instead we change this to a zero and we change this to a zero now we have a summary address that looks as follows we've got a 192.168.0.0.23 does everybody see that all we did is we went the opposite direction previously we were moving our network this way in this case we're moving our host bits this way so really all we did before is we took a slash 24 and we took a slash 24 and we aggregated them together into a slash 23. that's it so now everybody let's take these next two subnets 192 168.2.2024 and now this one for example is the 192 168.3.0 could we then because they're contiguous meaning one right after the other then switch this to a zero then switch this to a zero and now have a second slash 23 that looks like this 192.168.2.0 23. everybody see that too because all we did here is we did the following we took these two oops we took this address borrowed a network added a network bit we took this address added a network bit out of a out of the host bit i should say and that came up with a slash 23. now guess what we did over here we took this one over here took a whole extra host bet from the network and created another slash 20. now we've got two 23s does everybody see that so far because it's fundamental what we're doing ashley hughes we gotta know now right now if i tell other people i have access to subnets 192 192.168.0.0 and i tell somebody else that on my router i have access to the 192.168.2.0 subnet i've reduced the number of routes that i had to tell the rest of the world about to two but i can still reach all those subnets and why can i reach all those subnets because this is inside of this and this is inside of this now this subnet contains this subnet and this subnet since all you guys see that now i want to get more elegant can i now take this and this and make it even better go over direction one more can i now take dot zero 192.168.0.0 22 just moving it over which effectively makes my new mask look like this exactly joe white excellent excellent excellent excellent ashley excellent so that's all we're doing with summary addresses we're taking our subnets and at least from an address perspective we're aggregating them in something we can send out so carlos with f's we're dealing with ipv6 because ipv6 is in hex with binary and ip addressing ipv4 we're only using 0 to 255. there's no f's in binary now there isn't the mac address level because that's hex but not not with ipv4 so now let's do the same same thing two more times so now everybody ten dot zero dot zero slash twenty four ten dot zero dot one slash twenty four ten zero dot two dot slash twenty four and ten 10.0 324. can anybody think what we could do here could we roll up a slash 24 a slash 24 a slash 24 and another 24 into a slash what what would give us four of these subnets we just did it but somebody worked through this um okay so we don't need to make this scary so we can they're contiguous the slash 24 and a slash 24 combines into a slash 23 right okay now a slash 24 here in the slash 24 aggregates into a slash 23 right so we've got two slash 23s we want to go up one how do we aggregate us two slash 23s together what do we need okay we'll do it all manually one more time you guys look at it again so because you guys did it when i got it manually amrap you got it so if everybody else sees what amran sees excellent if you guys don't see what amaranth sees i'll let's work it through okay ashley you've got a slash 22. it's exactly a slash 22. daniel gets it it's the slash 22 excellent so now we're going to make it harder i'm not sure i completely understand that but we're going to make it harder we're going to add a level over here so here's how we're going to add a level now we're going to go from 10.0.0 all the way to 10.1.7.0 and guess what it's not going to be that complicated how many contiguous subnets is this one two three four this is the same one so let's get rid of that 5 6 7 8. how many contiguous subnets do we have over here we got eight of them right well two slash 24s turns into a slash 23. four slash 24s turns into a slash 22. so what's bigger let's double that size slash 21 amrap okay well amaranth got it it's going to be a slash 21. does that make sense to people pierre's got it um blackjack yep that's the answer so that's the answer um why does aws reserve a bunch of addresses for experimental uses um it's their choice they've chosen to do it and it's their system so some of the cloud providers do it as well but you can run your business any way you want but given those experimental addresses aren't really being used i don't think they're serving any purpose for anybody how about the 10-1 what about it the the 10-1 so the 10 the zero the one oh um that was actually great job there tom tom sometimes what happens is uh i draw my graphics and then i send them to someone who's much more artistic than me sometimes my graphics actually get drawn on a napkin sometimes they get drawn on a piece of paper or sometimes i'd take a picture of something so occasionally some of the people that helped me draw the pictures occasionally make a mistake or i gave them a mistake and they couldn't read my handwriting because they're getting you know doctor nurse practitioner handwriting which is pretty sloppy so i apologies for that these things just definitely happen um so the second accept so the question is a seam hun slash 15 so a slash 15 would give you two slash scenes put together two slash twenty twos uh two slash twenty four zika slash twenty three four slash twenty fours equals a slash twenty two eight slash twenty fours equals a slash twenty one sixteen slash is equal to slash 20. 32 24 is equal to slash 19. 64 slash 24s equals the slash 18 128 24s well now the math is getting a little goofy in my head but you get the concept that's the way you aggregate it it goes in reverse by a bit one at a time the same i hope that makes sense so now let's start putting the pieces together now that we're off some of this kind of goofy language here's where it all goes wrong everyone we're going to now talk about how it should be versus how it usually is the reason i make it clear and i say people should not use subnet calculators it's not because i have an aversion to science it's not because i have an aversion to tools at all i love tools i don't know how to use tools but i love tools i've got all kinds of hammers and screwdrivers from my friends come over because they impress me they know how to use them i wouldn't know at all oh the 10.1 so if you had a 10.0 and a 10.1 those two could be aggregated into a slash 15 on the scenes so yes that would be a slash 15 if you were going to do that the answer would be yes but you realistically speaking when you have to do this they have to be contiguous subnets so let's talk about a bad addressing scheme by the way this is what i see all the time this addressing scheme that i'm going to show you is one of the reasons that i will tell you that in in my life i have troubleshooted network after network this is what happens when someone doesn't understand subnetting but uses a subnet calculator they do this and by the way this will completely work until it falls apart let me move the stuff over so this is something you're going to see constantly in your career you're going to have somebody that's done this they gave the data center these subnets 10.0.0.2410.0.2 24 and 10.0.0.3.24 and they have 10.0.1 10.0.1.0 24 sitting in the cloud they've got that subnet for servers now if you've got four subnets this is great and you'll never have a problem but the companies i work with have 40 000 subnets at minimum and heck some of them are connecting the internet service providers and taking in three quarters of a million subnets from 10 different internet service providers so that's where these things break in this case we've only got four routes we only need these four subnets so plus whatever the wan link is so we've got nothing to worry about but you're gonna see a lot of this this is a disaster now what's going on here is right now you need to send these three routes from the data center to the cloud and the cloud provider will need to send this route over now i want you to think about this does anybody know how many routes you can send to aws from your data center total routes it will only tolerate a hundred 100 routes that's it that's all you can send to aws and it's not because aws isn't a great cloud provider and it's not because they don't have sophisticated routers they only tolerate 100 routes which is basically zero because the reason they can tolerate 100 routes is they've got millions and millions of customers and if they had to take in 50 000 routes from 10 million customers there's no routers in the world that can do it so they limit themselves to 100 routes so their systems can scale the routing takes cpu it takes memories so they limit how much you can give them to protect themselves otherwise we'd all be ddosing aws as network every single day by sending them standard networking so they can't handle it no cloud provider could handle that no routing protocol could truly handle millions and millions and millions and millions of customers with thousands of routes each so aws forces you to to cut it down to a hundred and how do you cut it down to a hundred you better be really good on your addressing scheme right now we've got to send three subnets just from here to the cloud now if this was addressed differently that wouldn't be the case now let's talk about doing it right see what you did here is you had the 10 address space here and you had it here which means i can't send a single summary route i can't send this route which is what i'd like to do i'd like to put i'd like to be able to send the cloud this route 10.0 and i'd like to be able to combine this 24 this 24 this 20 flare and make it a single route meaning a slash 22. why can't i really send a slash 22 over here because part of that slash 22 is sitting on the cloud because whoever set up the subnets didn't think about it ahead of time and they put part of this they should be here over here now technically it's still going to work right now but as if we scale this it's going to fall apart exactly we're sending we're almost sending the cloud a route to itself exactly pr now if we've got more specific because it'll be directly connected in the cloud so it'll temporarily work but we're going to have routing problems later now by comparison if we were smart about it this is the way we want to design it so look at it this way by doing this way we've got a 10.0 a 10. 10.0.1 a 10.0.2 and a 10.0.3 that all summarizes into this now the cloud has these subnets how many routes do i need to send to the cloud for this environment one one because once the cloud note that if the cloud knows to reach 10.x go to the data center the routers in the data center know how to reach everything but we only need to send a single route to the cloud now what if our data center had all these subnets what if it had just out of arguments like let's say we've got these subnets let's add some more subnets let's add a let's add a 10.0.4.0.24 this is going to get ugly but anyway maybe i don't want to do this maybe it's going to make it ugly but let's say i wanted to add 10 more let's say i wanted to add the 10.0.4.0 10.0.5.0. 10.0.89.0 10.1.3.0 10.2. or 10.2.3.0 10.3.3.0 10.4.3.0. i could still send a slash 10 route in fact i could send a test a 10 8 over here if i chose 2 to the cloud if i really wanted to i could send that to the cloud and then i could basically have 254 subnets here i could have 254 subnets here and i could have a bunch of subnets here so you see where i'm coming from we by addressing our our address is right we only need to send a single route or a summary route and then we can do some traffic engineering and determine where our traffic is going to go and when it's going to go there so these are kind of those things that we're trying to do and why we're trying to do it so let me try and see if i can show you a little bit more of what we're trying to submit here it's these things don't exactly completely come across perfectly we're trying to show them on youtube most of our content is really made for zoom because we we but you know when we try and do stuff for the community we try and do some stuff on youtube for it and the reason we choose youtube versus zoom is this way people can go back and see the recordings later so we want to really give the community a chance to be successful so that's what's going on here we can change this so now let's go play with it and let's go start thinking about how we would design the systems but i'm going to say it one more time for logical router addresses we use a slash 32 for wan links we use the slash 20 via slash 30. for what for land link so when link slash 30 loopback addresses on router 32. for lan links we use 24 but we can go as big as big as big as a slash 23 and that's about as big as a subnet wherever i want to get due to broadcast so i'm going to do this we're going to get in some architectural design with systems we're going to talk about laying out your systems and we're going to talk about optimizing the routing so we're going to get into the good stuff real soon but i want to make sure we address questions chris from my team is going to aggregate questions so let's make sure we do them first pierre do i always make sure to give a network prefix that is different in your land than the one from the cloud pr we're going to get to that soon but i always choose a different cider block for my lan or my data center than i would for the cloud and i always choose different cider blocks per cloud and i'm going to walk you through that so you can see exactly how and why great question by the way are there others chris no that that one was uh there before you asked for questions so okay so let's see if there's any other questions and if so we will answer them and if not we're going to move over to the fun part the architectural part the design part yeah this would be a good time for me to find out if the jeopardy theme song is public domain or if it's or if i have to pay for it that sounds good i always do [Music] with our 10 second delay that's really a great idea so let's see if anybody has any questions and if not what we're going to do is we're going to go have a party and we're going to do some network architectural designs first we'll do it from an iep addressing scheme and then if you guys want we'll add some bgp because you know there's me there's bgp you know i've got 10 000 hours of bgp i literally traveled the globe consulting on bgp and penn multicast and multi-protocol bgp so i love talking about bgp so we want to talk about some bgp and some traffic engineering after we go through the addresses and you guys are all having fun and you just say bgp we'll do some bgp so aws bgp cisco bgp it's all everywhere love talking about networking networking is the plumbing so it doesn't look like i see any more questions it looks like everything is doing great um i appreciate the ne um so tom i i know you've got it so effedestopoli and uh we'll keep moving forward so now um let's start getting into the architectural design which is one of my favorite components so let's now think about this now let's talk about why why why i'm really big on these things so let's go go in here we'll do a we're going to come up with a new slide i'll share my desktop so let's work over here so now let's say and then we're going to talk about how we can really really make this great so let's begin let's say we've got a data center here and data centers are here to stay even though the cloud the cloud is here to stay as well so now let's say we've got our data center now when it comes to real cloud architectures outside of solution architect certification exam but we actually do in the real world we're dealing with multiple clouds now typically speaking to to let you know how prevalent multiple clouds are a year ago 29 of all people were on a single cloud provider right now three percent of organizations that use a cloud or on a single cloud provider and everybody else is multi-home to multiple clouds so there is no such thing as a single cloud provider anymore that's why we teach multi-cloud architecture that's why we've taught multi-cloud architecture forever and the reality is a single cloud is a single point of failure so it's just bad architectural design in general so we're always going to be talking about multiple clouds and that's the reality of our lives so now let's look at it this way let's look at exactly how elegant we can get with our addressing and why we're going to do it so let's say this were aws and let's say this was azure it could be gcp it could be oracle it doesn't matter i'm just picking aws and azure you figure that's got to be what 85 90 of the markets are so now let's look at this over here and let's really make this elegant if azure is is using this block 172.16.0.0.16 and that's your main cider block and you've got 100 subnets in there who cares and in aws you're using the 172.17.0.0.16 and you're using over here 10.0.0.0.16. you need and you're connected and you've got a direct connection to aws you've got a direct connection to azure you've got a vpn backup to aws you've got a vpn backup to azure you've got a private line or two direct connection express connect whatever you want to call it it's the same concept between azure and aws now do you have high availability connectivity in this situation so we're going to evaluate the network from a couple perspectives well can you reach aws yes if the direct connection goes away can you reach aws cs if the direct connection to vpn connection goes away can you reach aws yes you can go through the azure direct connection or you can go through the azure vpn if the azure direct connection goes you still got a vpn so you've got a highly redundant environment in this so from networking right now this already checks all the boxes of a high availability system no single points of failure now inside of here let's see what network addresses need to be seen by soon aws and this cloud only needs to send a single route and the only route that aws needs to send to any of its neighbors is this that's the only route that it's going to need to send is this specific subnet it's going to have to send that route to azure and it's also going to have to send that route to the data center and we must know it on both sides i don't know what happened there bear with me bear with me we'll get this straightened out i think i bumped a button somewhere along the line all bear with me a second we'll clean this up for you a little bit so i can make it look as good as we potentially can it's okay so there should go that should work okay so nope still too big there's got to be better software for that we'll figure out for sharing files in real time but okay so what we're talking about over here is aws needs to send this route both to azure and we also need to know about in our data center now what will happen is these routes are going to be sent via bgp so if if azure tells the data center about it and aws tells the data center about it bgp will ultimately look in its routing table and it's going to prefer the path with the highest weight prefer the path with the highest local preference but for the one with the shortest day of path and to reach aws this link will always be preferred and this link will be a backup now at the same time azure is going to need to send routes to azure is going to need to send two routes to two places azure is going to and i don't know why this thing keeps happening every time i do some cutting and pasting azure is going to send need to send this route it's going to have to send it to aws or otherwise aws won't know how to reach it and it's also going to have to send it to us in the data center so what do we need to know in terms of how to do this hey mike just just share the powerpoint window in stream guard and i'll take care of it okay sounds good oh okay nevermind yeah it's it's sharing the window the whole window okay perfect okay so are you moving stuff around or was that me from before okay so while we're here i'm assuming everybody can see can see my screen what we're going to do is as follows note the number of routes that we're sending and where we're summing them so so we're sending this route to the data center and then what routes does this data center need to tell the data center needs to tell aws and azure about its routes and what are its routes i'm going to tell you right now i'm going to send the 10 dot 0 0 16 it's going to send it to two places it's going to send it to uh to aws and then it's going to send it to azure how many routes do we really have floating around our wide area network here when we really think about it this root this root and this root so there's only three routes in the routing tables that are being sent from three clouds only three roots just three of them now it doesn't matter how many subnets we have in here how many hosts do we have in the slash 16 65 000. how many hosts do we have here in the slash 16 65 000. how many hosts do we have here 65 000. how many subnets can we put in these ranges a whole lot a whole lot so azure is going to have to send the 172.16 to aws for example and aws is going to have to send the 178.217 to azure and azure is going to have to send the 17216 to aws and aws is going to send the 17217 to the data center our data centers are going to be sending the 10.0.0.16 to aws and azure and then in the end what we're going to get is three routes three routes three roots now why could we do three roots because this cider block is discrete and this cider block is discrete and this cider block is the street now if this was all in the 10.0.0 space and this was all in the 10.0 space now could this still work well if we change the subnets well the answer is possibly yes and probably so for example if the subnets that we were really using in this block we're and we're going to i'm going to make it up on the flies to bear with me if my mental math is wrong because i'm thinking about it i'm logically doing it but if what we did over here included these subnets 10.0.1.0 and this was let's not do that and then this was 10.0.2. and this was 10.0.3.0 and these are all slash 24s and then over here slash 24 slash 24 slash 24. now we did the same thing over here but now we had 10.0.4.0 and then we had 10.0.5.0 24. and 0.6.0 24. 10.0.7.0 24. now on the surface this is going to look really good but i'm going to show you why it's going to break and be a disaster which nobody thought of in a minute so now we'll go over here and because this is this is a good addressing plan um with no thought but mathematically this is a good routing and switching subnetting plan but i'll show you where this is going to break so let's go over here so now we've got eight 9 10 and 11. let's do this real quick let's get rid of let's go back over here now we go back over here let's get rid of this let's get rid of this because it's getting too busy it won't make any sense let's get rid of this let's get rid of this let's get rid of this let's get rid of this let's get rid of this let's get rid of this let's get rid of this now oops i think i just deleted something we needed okay let's just not get rid of that 172 16 because every time i try and delete that it seems to delete other things so right now this will work does everybody see how that even though we were all in the same cider range our addressing will still work because it will our addressing will still work in this particular environment and it's going to work great because we're still we don't have any overlapping ip addresses so it's going to work now where things are going to get ugly though with regards to it working are as follows what happens if we need an additional subnet over here we've used contiguous ones over here zero through three now this one has four to seven and now this has eight to eleven where do we add the dot for oh wait we can't so where's our next subnet going to start if we need a new one we're going to have to find something unique like 10.0.12.0 24. and then if we need a new one what are we going to do 10.0.13.0 24. and then what's going to go on over here oh wait we need a new one this is now going to be 10.0.14.24 and now we're going to go over here and we're going to get into this ugly range of 10.0.15.24.0 24. now you can see how rapidly this gets big and ugly because you can see that so that's why we need to think it out so that's why if you're using private ip addresses for your subnets or your vpcs and you're going to be using private addresses anyway pick an addressing plan that's going to give you room to grow maybe the 10 address space maybe the 170 216 for one vpc maybe the 217 for another vpc maybe the 172 18 for another vpc the point is make it so you can just send one route to your people and by sending just a single route to your people here's what we ultimately do we create an environment where we need less routes in the routing table less routes in the routing table means less memory in the routers it needs more simplicity more elegance on the ability to traffic engineer so now that we've got the fundamentals which we've just talked about which is the subnet and the supernet and we've talked about ways to spread your addresses out and by spreading our addresses out delicately in a matter we put ourselves in a position that we can design it so before we optimize the routing and we're going to get into some fun routing things i want all of you to optimize and design a system we're going to do it real time i'm going to ask you the questions i'm just going to be mike the ceo instead of mike the network architect all those years of working for cisco and training engineers and training engineers at cisco and riverstone and worldcom which is now rising we're going to do it you guys are going to teach me because i know you guys are all network architects cloud architects and experts and what's the point of doing all this training we're not based on certification we're based on getting cloud higher big difference anybody can get you certified we get you cloud hired so let's get your cloud hired in recording let's do some architecture work because we're cloud architects what do cloud architects do architecture works so let's do some architecture work now let's play with the network so let's have some fun here okay so now we got a new slide we're going to deal with a whole bunch of clouds let's say we're going to deal with aws let's say we're going to deal with azure all right let's deal with gcp okay so now because we're real and this is practicality this is the this is the data center all my students are become become knowledgeable on how to build private clouds because organizations use them so let's say the data center is running an open stock cloud everybody runs an open stock cloud american express runs emo on openstack cloud american airline runs an open style cloud um you name it verizon runs an open stock cloud ibm runs an openstack cloud lots of organizations using openstack cloud now we've got four organizations your task should you be able to complete the mission is to make sure that only one route is sent between every one of these vpcs we would like to assume that each cloud could get as many as 25 to 30 000 servers and we want to allow room for grow so assume we have 30 000 wormware servers but no more than 65 000 that is going to be in aws so say 40 000 aws 40 000 on azure 40 000 in gcp and you're going to use subnets that are going to give you less than 500 hosts per subnet but you might have to put 35 000 people to 40 000 people in aws 40 000 servers in azure and 40 000 servers in google and you know what in our data center you know what we also have 40 000 servers so i need you to get scalable i only want to see four routes in the routing table one route from aws one route from azure one route in google and one route for the data center somebody picking a dressing scheme we've got a class a class b class c addresses we can use we've got let's use rfc 1918 private addresses so what are those addresses rfc 1918 well we've got the the 10.0.08 we've got the 172 16 all the way to 172.31 16 and we've got the 192.168.16. so somebody out there come out there and tell me how to make my addresses because when you do this that's when we're going to build on this abigail marks give me a give me a something we're going to use for the data center pr okay the 10.2.0.0.16. it's an odd one to start with so pierre are you doing a 10 so if you want to use that for the data center and we could use that because it could potentially be a very good one pr what would you then use for azure or gcp and aws because maybe i missed one for you there because pierre your 10.2 is a perfect subnet to use i just uh i just want to uh see what your logic is we can totally use it all right if i have to lead it i'll lead it can anybody think of anything i'll tell you what i would do i'd work with pierre suggestion but it's a tuner a little bit i would make the data center or what we we we do i would make it a 10.0.0.16. what i would do i'm going to go to aws go to 10.0.0 i go to 10 got 1. at 0.0016. you know what i would do with azure nowhere is abigail at azure i would make this a 10.2 and over here for google i would make this a 10.3 i said we needed the ability to have at least at least 30 000 servers how many servers or could i potentially have in an environment where i have a slash 16 solid mass 2 to the 16th minus 2 somebody with a scientific calculator 2 to the 16th or better yet do it in your head 2 4 8 16 32 64 11 28 2 56 5 12 10 24 20 48 40 96 etc okay there you go it seemed better yeah you did it with the math smarter than me six five five three four great job essene so that's really what i was trying to impress upon you each one of your environments gives you 65 000 plus addresses amaranth great job and using the calculator is probably smarter than using your fingers like me but i do lots of stuff in my head so you know we're gut exactly with aws we've got 65 000 with 65 000 with azure 65 000 with google and 65 000 in our data center do we have room to grow is that enough everybody four roots nice and simple and elegant it's going to work really well um amelia olakomasta as well so you know these are great things and we're thrilled to have and we're thrilled to talk about it and that's why we do it that's why it's elegant put your things on boundaries now let's take this and let's make it look a little bit nicer now let's think about how now that we know what can we do how do we tune it how do we engineer it and how do we make it work um actually no caroline um if it was a slash 8 it would be 10 and then it would be 11 and then it would be 12 and then it would be 13 and then it'll be 14. um that's where the boundary would go so slash eights are where our things go from that would be go from ten to eleven a slash sixteen here's the place we'll be manipulating we'd be nippling the second octet the slash twenty-four we've manipulating the third octet so uh um no caroline it's actually going to be that we actually have it correct here but i really do appreciate the uh sanity check and trust me i make enough math errors that the saturday check is always a good thing to have so now let's look about why we're being careful with this so now we're talking about addressing and because of this we have clean routing we know to reach aws go here to reach 10.1 go to aws we know to reach 10.2 go to azure and we know to reach 10.3 go to google and we also know in the data center reach 10.x and now we're in the place everything's classes ips everything's smooth and running clean now let's look at the traffic engineering now we've talked so far today and as we're talking today we're talking about the subnet masks so as we're talking about subnet mask lens and as we're talking about these kind of environments now let's think about what we're talking about how far is the subnet mask so routers look at subnet masks the longer the subnet mask the greater the specificity of your routing information so let's get a little bit into routing because i want this to be a really good experience for you and i want to tie it back to things that matter for you so look at it this way if i give you this subnet and i give you this subnet okay 192.168.0.0.16. this is a big address block does everybody see that if i take this this address and i subnet it down i can get a 192 168.1.0 24 and a 192.168.2.0 24. and a 192.168.3.0.24 and a 192.168.4.0 etc etc pulled out of the slash 16. we've been doing this today but is that abundantly clear to everybody if that's abundantly clear to everybody type cloud hired in the box so i know you're here because like this is really important before we dive deeper i want to make sure you get that i want to make sure that you understand that all these things let me know what the cloud hired and get it as a subset exactly a subset exactly what's going on so now if i gave you two let's pretend you're a router today instead of being a person you're now a router i'm a router chris is a router daniel's a router mark's a router pierre's router tom's a router we're all routers today now if i gave you the information 192.168.1.0.24 and i also gave you a route 192.168.0.0.16 which is more specific which points you to something where you know exactly what it is do you know for example 192.168.1.0 does that identify it very specifically to the slash 24 these 254 hosts where this 192.168.0.0.16 could be any of 65 000 plus hosts is it clear to everybody that it's more specific the 192.168.1.0 so the longer your subnet mask is the more descriptive your information is so by complete by comparison one of these days i'm going to learn how to use windows or else i'm going to go back to a mac so now back to now that we go over here and if we go back to this 192 168.1.24 now which is more specific 192.168. 28 or 192.168.1.0 30. so let's say you you're you're a router and let's say you let me start with the slash 16 on the top so i don't drive everybody's body with uh the way it appears so we should be working around at least logical order so now going here everyone what is the most specific which gets you knowing what the subnet is within one or two host bits the slash 30 yogendra great caroline exactly so it's most specific because it is most descriptive now if i tell you explicit information are you more likely to believe it than if i give you more general information what if i said to you my cat cindy weighs 3.1 kilos and she eats 1427 kilocalories per day is that very specific about her weight and her dietary intake or what if i said my cat cindy weighs between 5 and 10 pounds and she eats a lot which is more specific giving you an exact weight and a number of kilocalories or kilojoules of energy we're just saying hey my furry cat you know eats a lot so the better your access to information the better decisions you can make so what routers do is they look for the most authoritative information and what is the most authoritative information the most specific my cat cindy weighs 3.14159 kilos which happens to be pi to a certain number of digits it's very specific other than my cat cindy weighs a lot today so routers take the most specific so this is your secret for everything this is your safe traffic engineering this is your secret for high availability this is your secret for everything so specificity of information so now let's deal with common environments you've got a data center happy data center i love data centers i've been building them for 30 years almost 30 years of data center because i love the cloud the cloud's another data center it's all let's say this is aws okay we're going to connect to aws okay let's do some architectural work i need to connect to aws i run latency sensitive applications where i need my latency to be consistently less than two milliseconds and i need at least eight gigabits of net of throughput per second okay what kind of connection do i want here everybody do i want a vpn or do i want to or do i want a private line or a direct connection and why somebody tell me in the chat box it seem i do need a direct connection alex i do need a director and the reason i need a direct connection everyone as i just told you i need consistent latency and i need consistent bandwidth so it has to be that because internet traffic is best effort when traffic goes through the internet we don't know what's going to go and our latency is going to be all over the place and we're going to have variations in latency and that's called jitter which is never ever ever a good thing so that's there so now as all of us understand these things now let's see what happens if you have two direct connections when you use a direct connection with aws do you use static routes or do you use dynamic routing with a vpn you can choose either but with a direct connection which is what pretty much every major company is going to need or multiple direct connections what's going to happen is they're going to need some consistency and when you use a direct connection you're going to have to use mike's favorite routing protocol bgp so bgp so when we're dealing with bgp bgp is going to help us determine the path through the system so no if you've got let's say you need a 10 gig low latency primary connection if you need low latency for your applications you're not going to get away with a vpn backup so you're going to have a direct connection and a direct connection backup now here's where things go wrong and here's where you need to know routing and if you know routing you can make the simplest most elegant beautiful solutions in the world and if you don't know routing here everything will fall apart so let's think about it this way does anybody understand what's going to happen here now when you buy high availability connections you always buy them across two service providers always you would never in good architectural design by two links or two pipes or two connections from the same service provider if you had a choice because if i buy them all through verizon and verizon is an incredibly good service provider if i buy them all through ntt another incredibly good service provider or if i buy them all through a t another good service provider or a central link it doesn't matter all service providers have outages every last one of them has outages so we buy one link on save verizon and another link on het or ntt we always buy across two links now can anybody think about the challenges of buying links from two phone companies one link is one millisecond somebody else's link is two milliseconds and that just that will cause all kinds of problems if we're not careful so if we were to get two direct connections to the cloud and we do nothing and the cloud is this is their cider range 10.0.0.0. 16 16 and let's say for example the company has a 10.00 0.0 24. and they have a 10 dot one let's make sure i got this right dot 0.1.0 slash twenty four okay so actually so let's set let's say these are the two addresses that we've used yeah okay let's say let's say we're using these two subnets now right now if we set up bgp and do nothing these three subnets are going to be set on the top link and on the bottom link and they're going to look identical on both sides of this equation and when the router is looking at the route the traffic and when we deal with bgp we have to send our routes in both directions i'm going to make this just one direction to make it look simple for you guys so right over here now the router is receiving this routing information on the bottom link and this routing information on the top link does anybody see a challenge here how does the router know which links to take routes are the same they're equal cost so now on the routing table we've got two equal cost routes now when we're dealing with bgp there's always an aggregate there's always an algorithm and there's always a thing and there's a reason we're not going to use a link aggregation group here because we bundled two links across two different service providers that have do different latencies we have a problem so if we're going to use a link aggregation group b3 collector we might have three links in the link aggregation group through verizon and then we're going to have to have a backup link aggregation group through say 18t and we're going to bundle all the 18 t-links together and all the verizon links together but if we're only going to deal with say a 10 gig link to the cloud direct connection and the 10 gig backup connection then if that's going to be enough we're going to have this now what we have is we've got two sets of equal cost routes so what will ultimately happen is we may be sending data to the cloud on this link through a t and then half of our traffic could be coming back through 18 and half of it could be coming back through verizon so what happens is packets get out of order some arrive on time some arrive later because 18 t might be faster than verizon so if 18 t is one millisecond of verizon is two milliseconds which is in the realm of tolerance um what can happen is you can get your packets twice as fast through verizon than 18 you get them out of order so whenever you're dealing with multiple connections to the internet or multiple connections to your cloud provider you really need to do something about it so you've got two options option one is which is basically what they teach at the aws advanced networking level which is like intro to networking they say just block this now the reality is in 25 years of networking i haven't met many people that thought it would be cool to buy a 10 to 100 000 connection per month and just block it for the heck of it my customers don't tell me that my customers want to use their backup so you know given that that's the reality and you know the aws advanced networking is so basic you know it wouldn't meet the needs of any customer we need to do things that are more ccie like to make better use of the system for our customers so what an organization would do is they're going to load chair across these now how do we do that well there's a lot of ways we can do it but now look at this very carefully for right now let's just send this subnet on the top the 10.0.0.0.24 and now on the bottom let's send the 10.0.1.0 subnet now the routers sitting at the data center have two routes in the routing table and they say hey wait to reach the servers in the subnet take the top link and then they say to reach the 10.0.1.0 subnet take the bottom link does that make sense to everybody right now if you know to hit the 10.0.1 subnet on the bottom link take it that way and the 10 and the 10.0.0 on the top link take the top link does this make sense to everybody let me know what the cloud hired if you get it if you don't get it let me know and then what we'll do is we'll answer some questions alex we temporarily have a single point of pharaoh but we're gonna have to fix that we are creating a problem along the way with our fix but everybody sees this okay so mark you got it b3 collector um pierre you got it so right now all we what we're doing pierre here and marla here let's go through this one more time routers always take the most specific route most specific route so if on the top i send a specific route to this router that's going to be say sitting over here and this router over here is going to have a routing table and in its routing table the router is going to say hey to reach 10.1 go out this link to reach 10.0 go out this link so the routers are building a map where does this information come from it's going to be put into bgp and bgp is a routing protocol which tells everybody how to reach their subnets so if we do this right now and you know what this name actually caught it very perfectly um alex wood right now the top link only is the the only the servers of the 10.0.0.24 sum that are reachable via the top link and the servers on the 10.0.1.0 subnet are reachable on the bottom link and the bottom link only so if this link goes away will anybody able to reach the 10.0.1.0 subnet it the answer is no and if this link were to go away as it stands right now this link because that's where the bgp routing is no one will be able to reach the 10.0.1 subnet so does anybody think this is a good a a good situation right now i don't we've got a problem so here's what we have to do amarath you are correct this is not a good situation and uh um yogendra what happens if the connection is down bgp is not going to switch at all right now we have a major problem but we also solved a major problem right now what we solved is we've load shared 50 50 of the traffic is going on the top link 50 of the traffic is going on the bottom link and we don't have any out of order packets because we've told the link switches to go but we do have a problem what happens if this link goes away or this link goes away half of our subnets are not reachable so now let's send a summary route or a non-specific or a non-preferred route on both the top link and bottom link we're going to put it into bgp and we're going to tell bgp to tell the world about it so by doing it this way by telling bgp to tell the world about it that changes everything so now let's put this summary route 10.0.0.0.16 on the top now guess what let's send that summary route on the bottom okay now everyone what is the best path to reach 10.0.0.0.24 top link or bottom link 10.0.0.20.0 24. pop that link pop the answer top or bottom into the chat box we're gonna go on the top specific link exactly things working perfectly now everybody tell me let's make sure you guys all get this first exactly because the top is the most specific pierre you got it no well you got it b3 collector you guys are doing great fantastic now before we get into the failovers and the high availability part and oh wait pr you're right the slash 16 is going to be ignored because it is not specific but the secret lies in that slash 16 for redundancy purple carl kenya excellent hey carl kenya you're now dealing with bgp concepts which most ccnas don't even know and this stuff doesn't even come into play to the ccnp slash cci level so carl kenya great job nice now which link is going to be used to reach 10.0.1.0 the specific link for 10.0.1.0 is not going to be at the top or is that going to be on the bottom bottom bottom exactly exactly so the bottom is going to be used now i have a fee seek i have a feeling you guys got it i have the feeling that you guys are really getting stuff that's kind of pretty cool excitement's building so now let's say this top link completely goes away it's gone the link is gone it's all dead there's no more link now how are all these subnets reachable is uh is or is there any way to get to aws for the 10.0.0.0.24 subnet is there anything are there any routes that would say hey i have a path to 10.0024 it might not be the most specific path but i still have something that tells you i know how to get there yogendra exactly it's part of the slash 16. pr exactly so here's what happens the data center now knows hey i can reach anything in the 10.0.0.16 space you go over here so what happens is by using a more specific route we can put a more specific we can traffic engineer our path but by having that summary or aggregate route if the more specific path goes away guess what we've got a backup path and we're still good so this is why we're doing this you guys are doing fantastic i'm really proud of you i'm really excited you guys are really really doing great so let's keep at it you guys are doing great so now that's pretending we're talking about bgp so bgp has a decision algorithm everyone even i want you to look it up right now i want you to go look up the bgp algorithm um because i'm going to ask you to use it so as we go through this together so it's going to be the bgp decision process i'll give you a link bgp decision process i'll give you a link so you can all see it sounds good so let's go find something from the cisco from the cisco web page how does bgp select the best path well actually let's make sure this is accurate because i see it on the internet let's make sure if the if the next hop don't consider it for the path that's true if the roots are the same okay so this these are this is a very good example of the bgp decision process i want you to all look at it so here's when we're dealing with it with bgp there's a bunch of attributes and if you guys want to learn about bgp i've done lots of bgp videos lots of pgp demos and there's plenty of information on that and if you guys want we'll probably do it but let's just say let's play with bgp very quickly for a few minutes so in this particular environment and in this exact particular environment let's talk about it so let's say you've got multiple so let's say we've got multiple links what is the first attribute if the next top is not available don't put it in the routing table that's going to get pretty complicated discussion of interior gateway protocols exterior gateway protocols and everywhere in between ernest we're thrilled to have you and ernest honestly you can always go back and watch the replay if you're having fun so please please we're thrilled to have you so here's what we're dealing with so in this particular environment let's make sure we're going to learn our routes from aws the same way let's say we've got two direct connections and let's say we wanted to turn we want to use let's say the top one is 18t let's label our wan connections on the top one we'll call it att and on the bottom next while we're at it we'll call it ntt nippon telcom is a very good global internet service provider lan provider so we'll say att and ntt because they've got similar things so okay we've got an american phone company and a japanese phone company okay so now we're dealing with this now let's say we're talking about the subnet 10.1 this particular subnet and let's say we're going to send this subnet on both links but let's say on the top link we change the weight of this route to three five five five five and on the bottom link the route is nvgp listed let's just make sure okay so this is the weight attribute that we've tuned so the routers let's assume the router receives the information it's got this route with a weight of 35555 and on this link the router gets sees the route and the route has a weight of 32768 so which route does the router send its traffic on the top link or the bottom link well let's look at that bgp decision algorithm that i sent you if the next hop isn't available don't put it in the routing table if the path is good prefer the path with the largest weight so everybody which link am i going to be taking the top link or the bottom link and why yogendra the top link why the top link somebody told me why the seam the top link exactly why the top link it prefers the largest weight okay good so now i'm going to change the top length of 10 000. a bottom link will be used because we're going to choose the path with the largest weight okay now guess what after weight what do we choose the largest local preference so what's next we're going to take the route we're going to pick the one with the largest local preference so now let's say we've got 200 on the top link 100 on the bottom link let's let's switch them around 200 on the bottom link 100 on the top link which path are we going to take somebody tell me which path are we going to take for this route the bottom link exactly okay so now let's play with it let's do some cool traffic engineering now let's load shares you guys got it we played with the weight we played with the local preference and don't worry we're going to play with the autonomous system path we're always going to have some fun here this is the best routing protocol in the world and it's really really it's terrific so let's look at this and now let's get a little more complicated so now let's put two subnets and let's do some load sharing on the top link we're going to put 10.0.0.0.24 with a weight of 200 and on the bottom link we're going to keep this with a weight of of 200 and then we're going to add a 10.0.0.0.24 with a weight of 100. okay now which link is being used for the 10.0.0 24 subnet and which link is going to be used for the 10.0.1 subnet so right now um we should be uh dealing with we should be using both links the question is which link we're going to use for which subnet i want you guys to tell me in the window below nice job there up here nice job there caroline woohoo this is well above uh ccnp level networking you guys are doing so how is it now caroline you're my student pierre am i my student so how is it we ramped up so fast because we work really solid on the fundamentals and when the fundamentals are strong and you get the fundamentals we can escalate really quick it's really hard to go when you don't have the fundamentals io um well um you're missing the numbers in there but i kind of think i know what you mean so the so there's that so okay good so now let's play with a little more so um if by comparison and some of these things are going to get a little little goofy like the origin code um for those of you in the cloud computing world you're probably not going to be redistributing and calculating your igp and agp and monopolying it so let's play with things that you're going to be more likely ready to tune what are you going to be really likely to tune the thing that you're going to be really likely to tune is as follows is the autonomous system path okay so here's what happens i'm going to map this out for you and then we're going to go back to this again because i want you guys to get this because honestly just knowledge of bgp is one of the most critical things you can you can know in cloud computing so the way the internet works is it looks like this there's going to be an as1 and there's going to be an as2 and let's say we've got an as3 and maybe we have an as for so what happens the way the internet is structured you've got a whole bunch of organizations and each one of these organizations is going to be called an autonomous system and what happens is each internet service provider is going to have question connections to other service providers and there's not going to be a rhyme or reason for the pattern it's going to look like something like this you know these guys are connected to these guys who are then connected to these guys who are then connected to these guys and i use the term guys loosely just a reference for a service provider so here's what will happen let's say you've got this subnet over here 1.1.1.1 and noaa um pierre an autonomous system means something other than under the control of an organization so aws would be an autonomous system google would be an autonomous system it's basically just an organization now aws is probably several autonomous systems but we're really talking about a company so what will happen is if if 1.1.1.1 is sitting on as5 and you're in the position of as6 over here and you do on the router you do a show ip root 1.1.1.0.24 here's what you're going to find in your router you're going to get a route to it and it's going to tell you to go here but what you're going to see in the routing table is it's going to look like this it's going to have it's going to show the route and the route's going to look like 1.1.1 dot zero what you're typically going to see is a b in front of it to show you that it's coming from bgp and then you'll you'll always have a mask and the mask will be in cider notation this is what you're going to see on the routing table and you're also going to see these attributes at least from the bgp if you do a show ipbgp you'll see a5 then you'll see a4 then you'll see a3 then you'll see a2 and then you'll see a a a1 and then what you'll and then your and then you'll see if you're sitting on over here on this route and this is what the routing table you're going to look at if you're here and you're looking for the route for here notice what you see you see the route you see the prefix you'll also see the weight whatever it is so let's say the weight is weight is three two seven six eight and let's say the local preference is a hundred because it's the default and you're gonna see all the autonomous systems that your traffic passed through so it's gonna show you all the different internet service providers your traffic passes through so now i want you to think about this let's pretend now there's two links on your as6 you could learn the route via two ways you can learn it via here like we just showed you this link to this link to this link to this link to this link back to this link you could learn it that way or you could learn the same route directly from a5 to a6 so if you're looking in your bgp table on a router you're going to see this you're going to have these two this is what's going to be in the routing information base on your router you're going to be learning the route via this path and you're going to be learning the route via this path so now let's look pretty carefully at this route the prefix length is the same they're both slash 24s so one's not more specific than the other the weight is identical so we're not right from the weight we don't know whether we want to go this way or this way then we'll look at the local preference the local preference is the same so we're not going to go this way or this way then we're going to look at the as path which as path is shorter one as hop going straight to as5 we're going through as5 as4 as3 as2 and as1 which one looks like a more direct path go straight from a6 to 85 or go from a6 to a1 to a2 to a3 to a4 to a5 clearly it's a shorter as path to go straight there caroline exactly go straight from a6 to a5 because it's got a shorter autonomous system path it's a single internet service provider versus one two three four five so shorter less internet service providers most likely better now if all these things were a hundred terabit connections and this thing from here to here was a 100 megabit connection it might be better to take the longer path but we can tune it and we can no no when we understand it but we've got to do the routing so now let's go back over to here now the next thing we're going to talk about is something called aspath prepending now this basically means we manipulate the path that the root has been come through so previously when we go back to this slide when we learn about the root here we've learned about it because it's traversed all these autonomous system paths which is really really great which is really exactly what we're trying to do but what if we just wanted to tune it so now let's get let's instead of playing the weight let's let's let's call this autonomous system one over here as1 the reason we're going to call it as1 is there's not going to be a lot of numbers we have to play with so it's going to look prettier so let's call this as1 on the top link the 10.24 we want that to take the top link do you know what the route's going to say on it it's going to say as1 why is it going to say as1 because we traversed as1 to get there and that's what it should be and now let's do the bottom let's say we want the 10.0.1.0 preferred on the bottom we're going to leave that by default is as1 now on this top link can we make that second route look ugly what if we did a as1 and what if we prepended or added an autonomous system we manually typed as1 again so when bgp learns the route we learn as1 for the top or as1 twice now that is us manually prepending or adding to the path an additional autonomous system to make the route look ugly now on the bottom what we would do is we would prepend the next one for the next subnet and by the time we do this look what happens can you eyes all see that this is now the shorter path for the top and this is now the shorter path for the bottom but this becomes the backup path and this becomes the backup path so prepending autonomous systems is part of the bgp rfc so we're allowed to do this this gives us a place for manipulation so the answer to that is yes and that's why it doesn't break anything bgp has been designed for this these are the the networking we call them the nerd knobs and we're the nerds that know how to leverage the nerd knobs to basically say need a little more traffic going out this way tune this go that way tune this go this way so this protocol was used for this kind of scalability tuneability because bgp was designed for this scalability and tunability it's used to connect to external entities so by connecting to external entities let's look at when we connect to aws they're not our company so we don't want to give them access to everything we need to give access to the necessary routes and only the necessary routes and bgp gives us this flexibility now there's lots of cool things that we can do in bgp for example we can tag a root in bgp add a community to it or add something to it and for you software developers this is going to make more sense to you but let's say for example we just wanted to tag the top right we wanted to tag a root with something maybe we called it a community we can put a community over here we can call it community cat we can call it the cat community i got cindy the cat she's really caller we'll call it the cat community and over here you know what we're gonna do we're gonna create another one called the dog community more of a cat person but you know i love dogs love cats but whatever i like anything with fur feathers so now let's do the same thing on the on the over here we'll tag this route with uh a cat community and then we'll tag this other route with a community called the dog community on cat community dog community whatever we call our routes whatever we want to call them and that what we can do is we can set a policy in our data center that every time it sees the root coming with tagged community change the local preference or change or raise their weight and every time you see a root coming with doug community lower the rate so this is where we start getting scalable as architects we design ways to basically mark the roots something and then we can create a policy on the far end to look at the marked roots and do something about it so this is where we're talking about communities so let's think about how elegant this is i can create a community that says i can give you my routes but you can't send the routes to anybody else it's called no export and aws sends that so what we're really dealing with kind of here is an exactly pair it's like storing a variable so when we're dealing with bgp we've got a lot of places to tune things so now now they're out now if you guys want we're here we're doing good you guys want to walk through on the difference between ebgp and ibgp and why organizations use both if you do type aws bgp in the chat box and and then i'll know that you guys look for or looking for some more networking stuff because if it's networking we're having fun i don't mind teaching some other stuff while we're all here i just want you guys to all have an incredible career so i hear from pierre he wants some more aws bgp if i get at least 20 people to type in aws bgp we'll deal with a little more bgp we'll give you guys some extra bgp skills when i make it you know i've dedicated four hours for you guys today and i'm willing to give it as much as it takes to really get you some lessons we've done some really good traffic um in terms we've gotten through we've gone through much more than i was hoping to do at the speed that we were able to do these things okay so there's a few of you guys um caroline's uh recruiting uh ronald mcrae i can do some more bgp i want to make sure there's enough of you the guys that want it so that i'm not just like shoving bgp knowledge down people's throats because mike likes bgp because i've been a network engineer since i think i think caroline's making a joke about you know meta is facebook so she's she's saying facebook needs to come to the session i think facebook probably does need to come to this section i will tell you this from experience as many years as i've been in networking if i know a hundred people that truly know bgp properly in the entire world it's probably a lot now i understand that i may know 100 there's probably about 5 000 that know it but it's that small of a knowledge base looks like we got a question while we're waiting on let's let's ask you the questions before we do there yeah okay so the question is is question on the isp connection to each other can we assume all uh big isps are fully connected to each other generally speaking yes and uh maybe some mom and pop isps are connected to only up to one mom and pop isps are always going to be connected to several isps it may only be two or three but they'll always be connected to several um one connection is a single point of failure they'll typically be connected to several but all the big people are pretty much paired with each other so the answer is yes and tom ortheus for real i've been a network architect for 25 years i've interviewed 4 000 networking professionals of the 4 000 people that i've interviewed i was able to hire eight of them eight of them actually knew routing and switching the rest of them honestly didn't know anything about networking they were certified but they were not knowledgeable it is really really really hard to find a good network engineer and it is even harder to find a good network architect and when you can find them they are worth their weight in gold so i gotta say networking is the foundation for everything literally when i did my ccie 20-some years ago about three million people had attempted the exam back when i was a two-day exam um the people that started the cisco process and my number is 7417 they actually started counting at 10 28 because it was 2 to the 10th so of millions of people that started the certification process prior to me 20 some years ago i was the 6400 guy to actually pass it so there was that and that was back when it was a two-day cci exam it's gotten much easier now that it's a single day exam and the pass rate's gone through the roof um now that you don't have to necessarily read 75 000 pages and take a two-day test still a strong test and it's great for everybody but things have changed a little bit so now that we're going back to some of this bgp content over here let's walk through it chris do you remember the thing that i was going to describe on bgp duh no oh ibgp and ebgpn oh yeah yeah yeah yeah yeah now i remember okay thank you i was about to say ibtp and egp ebtp but i was like i don't know if that's specific enough excellent now i can remember so let's work through this everyone i want you guys to truly understand this is what all architectures are going to look like for all networks all interior service providers um so pierre if you just know networking you're very you're totally differentiated i don't think you need that much more being competent itself is just a way to differentiate yourself mark so let's go deal with this here so let's look at a normal organization let's look at their systems they typically have a core of their network what is the core of their network it's going to be really high performance routers so let's say we've got a let's say you've got a core of the network and let's make sure in the core of their systems r2 we're going to make it a really small core so we've got room to blow this up for you guys we've got an r3 we've got an r4 now this is the way all well-designed systems in the world are designed these are the cisco and juniper recommendations so typically speaking you've got a core and your core has very fast really really really high performance routers routers are computers that just calculate and forward traffic all day long and these could be 100 gig links and you and there's probably multiple 100 gig links between these devices so what we're going to do is we're going to put multiple 100 gig links here maybe in some link aggregation groups things like port channel ether toronto whatever term you choose to call it then because it's the core we want to make sure you've got ultimate availability so we typically speaking we'll fully match our core to make sure that we don't have any single points of failure now typically speaking this is considered the core of a network this is where your performance is you've got nothing here other than shape size and position i think we can send this to the back position well maybe not anyway so that so that's the core of our system and then typically what we have is we typically have another set of routers switches and this is going to be called the distribution later and then typically speaking these distribution layer switch routers will typically aggregate into two locations so and then what we'll have and i'll show you why we do this so that you understand how we promote high availability and and why we do this and then you typically have things called an access layer and what we'll do is we'll feed the access layer switches into our distribution layer switches which then get fed into the core now here's why we do this when you're making a three-tier web architecture you're decoupling it right why are you decoupling things why do we ever decouple things because if anything happens here we don't want the stuff here to know about it so by creating an access layer if something happens or goes goes haywire in this one switch chances are it won't affect this switch or this switch but if it does which it shouldn't it'll be blocked at the distribution switch and if it doesn't get blocked at the distribution switch it'll be blocked before it it'll it hopefully is isolated here if not it's going to be isolated here it's less likely to affect here and here so when we're dealing with our network architecture we're dealing with the decoupling wait web architecture load balancer web servers load balancer app servers queuing system database maybe some caching maybe some read requires why are we doing all this decoupling because it enables us to do one thing in one part of our systems and whatever we do in one part of our systems works in another part of our systems cindy you want to come up here so anyway so you know that's why we're doing these things that's why we're decoupling that's when we're modularizing for scalability so now let's look at what this is going to look like typically speaking before we get too deep we have a really important question that we need to address let's answer it let's answer it what is the difference between a switch and a router so noelle this is a great question and in today's world it's very blurry in real and the world where i come from it's not so blurry so a router operates at layer three and what do i mean by layer three a router basically goes from subnet to subnet to subnet so everything on a router router has two arms one subnet another subnet and routers route between subnets switches by comparison are layer two devices they operate at the data they data link layer so a router forwards based on subnet and a switch forwards based upon mac addresses so switches operate at layer two routers operate at layer three so when a packet hits a router and this is pretty important oh well the router does a lookup and it says destined to this subnet go out this interface and if it doesn't have a route in the routing table noelle you know what happens the router just drops the packet and it dies and that's what's supposed to happen by comparison a switch is a layer 2 device so a switch forwards on mac addresses so switches um have no rounding they have no ip addresses and everybody in the switch or the at least the virtual switch can talk to each other where every interface on a router is logically isolated and you have to route between subnets so when you're dealing with the switch and i always get asked this question so i always keep a switch directly next to my desktop i'm going to rest it in my head for a minute what happens is let's say you've got a a let's pick one of these ports let's say you've got a user here the router says who has the mac address for ip address 192.168.1.1 flood it out every port in this switch and then when this when this guy over here says i have it what'll happen is it'll put address in the mac address table the switch and it will send data only out the port that's destined for that switch and nothing else so switches operate at layer two and switch and and routers operate at layer three so switches forward based upon mac addresses stretch my neck out after resting some metal on it and routers operate based upon ip addresses now noelle when i say switch what happened was first there was the concept of routers that worked on subnets and then there were the kind of concept on switches that worked only on layer two and then about 20 years ago extreme networks was probably the first people to do it they came up with the concept of a layer three switch was a switch that could do routing and switching all in the same device and the way this would work would basically be on a switch you can group ports together let's go back to the switch over here let's say i wanted to virtualize this switch and i wanted to turn the switch and say all these ports exist in one virtual switch and then these ports exist in another virtual switch what i could then do in these layer three switches which are really routers is i could literally spay all these eight ports are all in this one ip subnet and all these eight ports are in another ip subnet and all these ports are another subnet and then route between them so the practicality noel in today's world of a switch in a router isn't that much different routers operate at layer three switches operate at layer two so that's the only main difference but that is a really good question so let's go back to this and honestly that was a really great question noelle often these things these access devices are layer 2 switches like that little pizza box that i had on my head and the distribution layer switches are typically routers or high performance switches that do lots of routing and what happens is the routers stop the broadcasts because they limit the broadcast and the routers enable you to do access control lists access control lists enable you to keep traffic out of your subnet and routers enable you to traffic engineer your traffic and filter your traffic and do firewalls and things that are much more secure so routers are basically what we're going to be using at layer 3. if there's an ip address it's a router if it's a mac address it's a switch and that's why they're starting to be used interchangeably but they're not technically so now looks like this so typically speaking an organization is going to have what's called an interior gateway protocol so let's say they're using something called ospf it's probably the most common interior gateway protocol used in today's world what will happen is all these routers are going to identify they're going to send they're going to tell everybody to exist they're going to be like hello hello hello and basically these routers will multicast to 224.0.0.5 and 224.0.06. i'm here i'm here wake up i'm here i'm here wait have you ever heard of this before like a keep alive and dns or a keep alive as in a load balancer where do you think it all comes from it all comes from networking and heartbeats years ago i'm here are you there i'm here are you there i'm here are you there so these routers constantly identify each other they're constantly saying i'm here if one of the routers disappears they remove the routes from the routing table that's they're like a gps for your traffic they consistently direct your traffic go here go here go here and your traffic gets to this location so here we do wrap we do routing where if this link fails this link's going to recalculate in seconds because it's internal we want to recalculate fast so here's the thing if i want to do math 2 plus 2 plus 2 plus 2 plus 2 it's either 2 4 6 8 10 12 14 16 18 20 22. now ask me to do advanced statistics or calculus at that speed not happening the more intelligent we need the more control we need the more tuning we need the slower it gets why do you use network load balancers because they're fast why are they fast they don't do a lot they look at the tcp udp header the destination protocol and port number and that's it why don't we use application load balancers because they're smart they're tunable but they're slow so it's the same thing the more intelligence we need this more slow it's going to be the less intelligence we need the speed in an interior gateway protocol internally to organization we're optimized for speed and self-healing when we connect to external entities like aws we optimize our protocols for tunability security and control so i hope that makes sense so now let's look at what it's going to look like what do these organizations look like how do they build their systems let's have a party with it so what you'll typically have is you're going to have your organization and inside of your organization let's fill it up with no colors so this is going to be you this is going to be the data center maybe the cloud i don't care cloud's no more than a virtualized data center and if we had a good router that we could use in the cloud we'd be doing the same thing oh so we've got our happy data center over here now let's say we've got an organization and let's say the organization needs really high performance internet reality let's say there's the internet let's think about this if we really want good internet routing there's about 10 internet service providers that cover about 80 percent of all the internet's customers 10 of them now what that means is if i connect to a t and a customer's website's on a tnt if i'm on a t and the customer's website's on a t guess what it's not very far for me to reach the customer now if i'm on 18 the customer's website is on verizon i need to take my systems to get to at t and then somehow i need to get to verizon well either that goes by me connecting to att who's connected to verizon or maybe i jump on att and they pass me off to sanctuary link and then they pass me off to verizon that's the quickest way on and off the network that could be could be you know the way my traffic goes but we don't really know but i'll tell you this the more internet service providers your organization connects to the less like the more likely you are to be within one hop so when you're dealing with really high performance web apps if you want to minimize the latency if i'm on the ntt network as it stands i've already got the country of japan covered because i'm on their local network for example and content delivery networks help mitigate this a little bit but if we're really talking about pure routing and dynamic changes if i'm connected to a t i get access to 18 t's customers on network if i'm connected to verizon i get them if i'm connected to vodafone i'm getting them if i'm connected to centurylink i'm getting them by the time i'm done i connect to singapore telco bt and a few other major telcos i'm connected to the whole world and my internet performance is almost as good as my private network performance because everything is on network not traversing the internet the same way so it would be pretty common to connect to multiple internet service providers so let's say we are a major ecommerce site and we've got 10 10 gig connections to the internet all across different internet service providers no big deal lots of companies would do this now do we think it's a good idea to put all of our internet connections on a single router all 10 connections all in a single router even if we're smart even if that router has multiple power supplies and multiple control modules and multiple line cards what happens if we've got all all everything we have connected to a single router a single 500 000 router that's connected to 10 internet service providers and something happens to the router can any can anybody tell me what happens to our internet connections if the router dies somebody tell me what's going to happen if the router dies in the chat box i promise you it's not going to be a great situation and nobody's going to be happy single router connected to 100 service providers if noelle exactly everything goes down if the router goes down everything goes down so not good so we think maybe we should split it among two routers everybody instead of one router we need to split it across two routers so now we've got router one over here let's call that's got five internet connections across different internet service providers exactly bernard we got a single point of failure i don't like single points of failure no single cloud providers no single internet service providers no single points of failure that is not a way to design a system for anybody anywhere anytime so now let's do it over here let's let's do this okay so now we've got the internet over here we've got router 1 over here we've got router 2 over here each one of these guys is connecting to an external entity on the internet what is an external entity an external autonomous system if we connect to an external autonomous system external autonomous system is this external bgp or is this internal bgp if we connect to external systems so pierre exactly noelle exactly when we connect to these external systems and connect to external systems it's external bgp so now i want everybody to kind of think about this the routes that are learned on router one router one is connected to five internet service providers router two is also connected to five internet service providers as it stands right now does router 1 know any of the information that's on router 2 does router 2 know any information that's on router 1 like for example what if the best path to go cloud careers what if the pat the best path to go cloudcareers.com is through router 2 and this internet service connection specifically will router 1 know how to do that not yet exactly be through collector it has no idea router 1 has no knowledge of the routes on router 2. router 2 learned it via external bgp now i'm sure you got two routers in your network and they're running ospf so they can reach each other but router 1 doesn't know router 2's roots and router 2 doesn't know the routes that were learned by router 1. so what do we need to do between these devices not only do we have to have network connectivity we have to run internal bgp so what is the point of internal bgp internal bgp carries your internet routing information to other routes inside your system so by doing this if your if a user's data gets sent and it hits router 1 but router 2 has the absolute best path to it router 1 will know to send the traffic to router 2 on the way to go cloudcareers.com by comparison if router 1 has the best path to it the traffic would come up and even if it hit router 2 router 2 would hot potato the traffic over to router 1 and then router 1 would send it to go cloud careers so what we're really doing these things is we're putting them in here and we're exchanging roots and we need that internal bgp to do it because we're going to be taking in a lot of routes we're going to be taking in like three quarters of a million routes or more from every internet service provider we deal with so we're going to get millions and millions and millions of routes that's what we're really talking about so we're talking about some really really cool environments and things that we can actually be dealing with to kind of let you know about it and that's why we do it so that's why we're dealing with ebgp to connect to external entities and ibhp to carry the load across our autonomous systems so why now if we just had a single router now let's say we've got seven and a half million routes sitting on these routers now could we just dump it into ospf through something called route redistribution and then teach ospf or eigrp or intermediate systems intermediate systems all these routes well if we did what happens is we would break things because it's not that we couldn't take routing information from bgp and to put it into our interior gateway product like ospf the problem is the interior gateway particles are not designed for this many routes and we can't filter things so the point being is um we actually have to choose a situation and choose a routing protocol for our needs so with any other part of our architecture here's what we're dealing with what is right for us at the right time that's really what we're talking about so i hope that's kind of clear for you guys so that's what we're dealing with ebgp and ibcp so now what do these systems really look like what does an inner service provider look like well an internet service provider might have two routers over here like we're dealing with they are that are connected to 10 isps then they may have their networks themselves and inside of their networks they've got two challenges well let's talk about the bgp challenges so let's let's deal with this here's the last thing i want to talk about bgp and why am i talking about this for bgp well if you understand this with regards to bgp you're always going to know what to do with regards to vpc pairing and things like this and you're always going to have head and shoulders and years and years and years advancement over everybody else so let's get you here so let's go here now normally speaking this is ebgp so you've got your ebgp speaker and your ebgp speaker is connected to your isp so this is ebgp on this link now over here this is your router that's connected to the isp now if router 1 tells router 2 and it will over here router 2 it's ibtp routes and if everybody in the company knew let's say they had a default route and the default route is a route that looks like this 0.0.0.0.32 that's your default route that says if you don't know where to go go to router 2 and router 2 and router 1 have ibgp communications between them for example what you ultimately have would have is you'd have your traffic here for the default route it would go to router 2 router 2 will have learned the routes via router 1 and it will send it to its ultimate destination and everything works perfectly perfectly perfectly now where are things going to break and i'm going to show you where things are going to break and they're going to break real fast and real quick and real easy so let's unbreak them now let's pretend we have this let's say we've got router 3. and let's say with regards to router 3 we decided it would be fun and a great idea to run ibgp in addition between between here and here and between here and here but ibgp is not transitive so now let's pretend this route this default route was in vgp but it's not and it was sent to this router but it's not over here at router 3. now let's say that we've learned all these cool routes the routes are passed via ebgp to router 1 and then the routes are passed to router 2. but they're not passed to router 3. why does router 2 not pass router 1's routes to router 3 ibgp is non-transitive that means the routes won't be sent more than one location away which means router 3 will never know about any of the routes learned from the internet so with ibgp it's not transitive so how do we fix non-transitive routing everybody well we do two things option one is we fully mesh our ibgp pairs so now we do an ibgp connection between router 3 and router 1. so now if router 3 needs to reach router 1 it knows what to do and it also knows how to reach router 2 so we fully mesh them so this is going to work perfectly perfectly perfectly as it stands right now because we fully meshed them what do we do with vpc peering we have two options we fully mash them or we do something else now i'm going to show you where this something else comes from so when network architects like me will tell you that transit gateway and cloudhub are the same and you deal with the aws people that are like that are not networking people and they're going to look at them as so different i'm going to show you why they're really just identical i mean they're all just a bgp route reflector the same 30 year old technology and it's great technology and it works and we've used it everywhere so what ultimately hebson can i don't know what you mean by mlsp if you're referring to a multi-protocol label switching mpls that's something entirely different but that's something different we're talking about btp right now i can and will do an mpls day if anybody wants me to in another day um but now we're really getting out of ip routing and we're now getting into label switching and a lot of that mpls stuff is going to go away when we're dealing with cloud computing their cloud providers are all going to be running it internally so let's keep you guys you know but if you guys want i'm happy to do an mpls day i worked on mpls for a quarter of a century and always thrilled to talk about it so now let's bring it back so if we have this if we have isp let's make it look a little more elegant for you so now it's going to look more familiar so now we've got over here let's call it our router now let's say over here got another organization let's call it a router 2. let's make a router 3 r3 r4 now in this particular um in this particular information um let's look at what we actually have going on here so now let's assume we have connectivity between these organizations now let's go back to our transit of routing we've got another router over here this router connects to the internet r1 let's call this let's call this r25 um numbering was goofy now let's connect the internet so r25 let's call it the internet let's say r25 um has two connections that are bgp now we're probably not going to put them all on a single router but let's just pretend we are for right now now what you have because this is our router and that's the internet that will always be considered to be ebgp now if we send these routes over here via ibgp in a perfect world this router over here this r1 would send the internet routes to r2 and r3 and r4 and if that occurred r2 would know to go to r1 to r25 out to the internet and r2 would know how to go out to the internet and r3 would know how to reach r2 through r1 if the routing information was passed from here to here to here so that's realistically what's working wrong so is the networking needs to go from here to here so so now let's think about this um completely um if router one was told take router two's routes and share them with router three and router one said take router fours routes and share them with router two and everybody shared everybody's routes everybody would be talking but ibgp as part of its loop avoidance strategy and non-transit strategy doesn't allow transitive routing and what does that mean it means that the routes learned that are pushed to our router one won't be sent to router 2. and it means that if router 2 tells router 1 about some routes via ibcp router 1 won't tell router 25 and it won't tell router 3 so nobody knows about anything that's why it's non-transitive routing now with bgp we can fix that we just call this device router one we call it a route reflector and then we specify all these devices as route reflector clients and by doing that what we're really doing is we're telling router one to take router's two's routes and pass them to router three and take the routes learned by router 25 and send them to everybody else and now once we turn this into a route reflector but we have we've got transit of routing no longer have to fully mesh our bgp pairs so now let's change everything let's get rid of this internet service provider we don't need it let's get rid of this ebgp pairing to the internet we don't need it now you're on the cloud this is your vpc price good everything in your vpc is working perfectly you got your web apps you get your business apps your erp systems everything's going good business is great everybody's making money life is good everybody's fine and now you decide that you're going to purchase three other companies they're all one company and this new company that you buy has a new york office a san francisco office and a london office and guess what you just want to connect them to your vpc so now we've got new york i've got san francisco or san jose san jose and now let's say london you all connected him to your vpc now guess what new york sends its routes to the awsv pc the aws vpc in new york can talk to each other san jose sends you its routes you send san jose your routes you can talk to san jose and life's good now london sends you their routes you send london your routes life is good we're having a party time everybody party at mike's house i'm now talking to the folks in new york i'm talking to people in san jose and i'm talking to people in london everything's great right well what if new york has to talk to london they can't what if new york has to talk to san jose they can why can't they we have non-transit around what does non-transit routing mean it means the routes that are set from new york to here aren't set to san jose so what are options we have two options here well it's the same two options we have with bgp the identical two options nothing changes at all so i'll show you what i mean here's the here's option one option one is simply do this let's get everybody talking to everybody we've got four let's say we've got four vpcs and you want everybody to talk to everybody we can what's called fully match them so let's say we took the vpc here the vpc here this is option one click connect appear here appear here peer here oops peer here peer here now everybody can talk to everybody because we're fully messed now this is great for ford devices but what happens if we've got a lot of devices so let's break this down we're dealing with this formula and the formula is going to be called n times n minus 1 divided by 2. so that means if you've got three vpcs that need to be paired or three ib gp pairs it's the same thing that's going to be 3 times 3 minus 2 or 3 times 3 minus 1 which is 2 so that gives us 6 divided by 3 is 3 period connections now we've got four four times three is twelve divided by two is six pairing connections but now we've got ten ten times nine divided by two is forty five as you can see it gets very big very fast and we can only deal with 125 vpc pairing sessions of aws so you're not going to be able to do that many so in bgp we have this concept of a route reflector if this is the route reflector all the routes that are sent here will be sent here which will be sent here in the cloud we have the same concept of a route reflector we don't call it a route reflector we've got this concept called cloudhub so what is cloudhub if new york has a vpn in and san francisco has a vpn in london as a vpn we terminate the vpns on cloudhub we tell cloudhub basically you're now a route reflector and we tell cloudhub to reflect the routes learned from new york to san jose back to london and vice versa everybody's routes are sent to everybody by reflecting everybody's routes everybody's talking to each other and we've taken non-transit of routing and called the transit planning so in the network world here's what we call it it's called a route reflector in aws we call it cloudhub then aws looked at it and here's what they said not everybody can get away with a vpn connection what if we need guaranteed performance guaranteed throughput then we need a private line so now how would we use a private line we'd use a private line in the following manners we would for example use something called instead of cloud hub we're going to call it transit gateway what is the difference between cloud hub and transit gateway not a whole heck of a lot it's pretty much the same thing but transit gateway works with private lines and vpn connections so realistically speaking that's what we're talking about we're talking about the same thing the same concept and that's why i want to make sure that everybody truly gets us it's the same concept so now who has questions um does anybody have any questions for me right now because we've covered a lot we started with ip addresses after we went from ip addresses we talked about ip addressing then we talked about transware gateway then we talked about cloudhub all of these things are aws services since we can't go there pr and then put one of our vpn concentrators there and we can't put one of our router's things there we're now in a position where everything is virtual everything is virtual so since you guys asked you got about 20 minutes let's do it it'll round out your networking knowledge i don't know how we did this at a subnetting workshop but you guys were all learning good you guys were all fast so we can talk a little bit about what is mpls so let's do it so what is mpls normally speaking you've got a device that comes into your network and this is what your routers do your routers have a bunch of interfaces mike can you confirm the formula again it's n times n minus 1 divided by 2. okay sorry some someone asked for just to verify so okay sure so n times n minus one divided by two okay thank you you got it so going back to this particular environment a router's got multiple interfaces so it's going to have an interface over here and it's going to connect to another router and i have a subnet over here this router is going to have a lan segment on it which you know generally speaking we draw like this and then let's say on the server let's say we've got a server that's got an ip address of 1.2.3. that's the server's address and let's say we're over here we're popping here at 192.168.1.1 which makes the router's ip address which is our default gateway on the same subnet a 1.0 let's say that over here i don't want to really think about anything too complicated let's say we've got 10 let's just one dot zero dot zero one slash 30. that's being used on one side which then means the dot two is going to be used on the other side what will ultimately happen is when this packet gets into this router and the goal is to reach 1.2.3.4 if this is your system your system is going to say do i have a route to 1.2.3.4 and your computer has a route table and it's going to look at a trout table and it's going to say no but i have a default gateway my default gateway is 192.168.1.0 so you send your packets directly over to here that's where you're sending your data so you send your data to the router and the routers here and the router now says the the packet then looks up and the router says do i have a route to 1.2.3.4 and the router says i do go out this interface so then the router sends the packet out this interface on this one link and now the packet sits in this router and this router says hey do i have a path to 1.2.3.4 and the router says in my routing table i have a directly connected subnet 1.2.3.0 so what happens the router sends that packet out this interface and the router will then send what's called an art broadcast which is who has the mac address for 1.2.3.4 and then what will ultimately happen is 1.2.3.4 will reply to the arp and it will say i have the mac address and then 1.2.3.4 is going to answer and it's going to then say do i have a route to 192.168.1.1 to answer mike's query and it's going to look in its routing table and it's going to say my default gateway it is 1.2.3.1 it's dot 24 and it's going to send its packet to the router now the router is going to look in its routing table and say do i have a route to 192.168.1.0 and it's going to say yes go out this interface and when the router's here and it's the router is going to say do i have a route to 192.168.1.0 and it's going to say yes i do and it's going to send an art broadcast who has the mac address of 192.168.1.0 it's going to send it to this guy this guy is going to respond and then a session is going to be established then our traffic will go be bidirectional and that's how it's going to occur note all of this happened based on destination subnet mask so destination subnet mask that's it we routed everything based on subnets so what is mpls instead of routing on subnets let's say let's say we've got four routers router one letter two router three router four so in this particular environment we build our four routers we create our links in between our routers and we enable our interior gateway protocol just like before let's say we're running our ospf and now we turn this into our own sort of like cloud so let's uh let's fill this we'll send it to the back there's a way to send this to the back right send it back okay okay so now in our mpls network now we've got an environment so first we turn on our ospf our interior gateway protocol then instead of switching packets we turn on a label distribution protocol or a tag distribution protocol it's another routing protocol and in this case i'm just going to type tdp tdp ldp there's lots of protocols that were used for this and then we're going to have to signal these things we're going to have to we're going to create a series of tunnels we're going to use something called rsvp and then what we're really going to do is we're going to come up with a primary path which is going to look like this we're going to explicitly tell our network this is the best way to go through on the network and then we're going to create a backup path and we're going to call and say this is the backup path of the network so now what happens we're going to send our data from this r1 it's going to hit the cloud but when it hits the cloud the cloud's going to smack a label on it so normally when we talk about 802.1q tagging we talked about smacking an 802.1q tag on it here we're going to smack a label on it so we're going to take our packet we're going to pop a label on it that routers are then going to switch based on labels they're going to stick this packet into this traffic engineered tunnel which is then going to go to router 3 router 3 is going to see the packet and say got it but it's now destined to an ip subnet the route that router 3 will then strip off the label and send it out its port to go on its merry way so what is mpls really instead of switching based on packets and the destination packet we just switch based on labels what is a label if i take my phone over here and i put my phone inside of a box and i ship the box via fedex and the person on the far end takes the phone out of the box and then picks up the phone and uses it that's what's happening when we use an mpls network all we really do is we take the packets we smack a label on them we then switch the packets from switch to switch based upon the labels and then when they get out of our systems we pull the labels off it's like taking our phone out of the box and then we place our phone call so that's really all that we're talking about over there so you know that was way out of scope for this but you know people asked the questions and i wanted to see what we could do to make everybody happy i want to make sure everybody truly gets a truly amazing experience so it's 4 38 i said i would run till 5. so do you guys have any questions if you do let's answer your questions because i want to make sure we give you a good experience so trying to bear with me so a seam can we have certain locations reflecting the routes while other regulations refrain from it um could you set up some devices to be route reflectors and clients and others not the answer is absolutely yes and you definitely would as seam hands also in a specific router location can we control the reflection of certain subnet blocks while some would allow others so either you're going to be a route reflector client or not now anytime you're dealing with any kind of peering sessions you've got a lot of flexibility distribute lists route maps filters all kinds of ways that you can actually tune what you actually send and wear so the answer is you've got lots of flexibility and policy-based routing you could be doing with bgp i mean i could fill up books without all the tuning you could actually do here i said i was switching using the tag how would it be switched otherwise okay so typically speaking going through the system normally we're going to switch based upon destination ip address router is going to look at the destination subnet and it's going to send it that way in this case we're switching based upon labels really it's just an encapsulation and all of what we do is really an encapsulation we want to send either knight signaling over mcls we're just taking one thing sticking it in something else and taking it out the far end so gre tunneling take a packet stick it inside of a tunnel ip6 tunnel stick it inside of an eyepiece tunnel any of these things we're just switching based upon another path so that's really all we're talking about but great question there pr truly truly truly great question mpls is just another form of encapsulation and d encapsulation it's all it is it just retakes an ip network to do it but that's all it is okay wonderful pair i'm so happy you got that perfect the scene assume i'm very happy you got it and that's why we're doing these things i truly truly truly want to get you guys to do it so while people are asking final questions if you guys had a good time if you can hit the uh the like button um it kind of helps signal the algorithms to let other people know we're doing a good job and we really try to work hard to provide as much training as we can for the community actually could you let us know where you're from i always love to see where you guys are from let us know where you guys are from in the chat box i find it truly amazing sometime it's because we've got some people that i've seen from all over the world alonzo my good friend i'm good to see you here i'm tom i'm curious where you're from you look like you've got a similar name to me so i'm always kind of curious um alonzo you're in katy texas tom in the netherlands fantastic pierre in montreal that's wonderful b3 collector pierre i'm so happy alonzo tom excellent um kenya i mean this is really wonderful montreal ireland sydney oh this is just so nice i wonder what time it is raleigh north carolina kenya from the south 101 from the south i know something about being the south i'm in south florida caroline raleigh but from kenya that's really great omar for rook austin test as well this is just so nice um amrat i know you're in toronto over there i seem as in india that's great bernard you're from cameroon in virginia that's fantastic kenya's from atlanta but you're going to go back and watch it a few times you know what in the beginning 20 some years ago i think it took me about 10 times going through the subnetting and you know what i got it so whatever it takes to get it is just an important concept and i try and run these periodically because i think it's so important i love the sunglasses that gun so thank you all honestly it's been a true honor and a true privilege spending the afternoon with you if you have any other questions let me know now otherwise on monday please please please register chris from my team is gonna gonna pop it there um so happy uh caroline and kenya carl if it takes a few times it takes a few times um evo and i know that's your name but i've been calling you that for a long time and i think you're amazing so i'm so thrilled you were able to participate any last questions for me otherwise please join us on the monday call my team's got a couple more videos since everything we do about is about getting you cloud hired i'm going to give you some special special special video that's coming out tomorrow this can improve your chances of getting hired everything we do is about helping you get cloud hired so i want to thank you all for this great time we spent together and i look forward to seeing you in another video andrew so nice to see you alonzo uh always appreciate it fantastic everyone take care everyone see you very soon

Channel: Go Cloud Architects

Views: 5,396

Keywords: networking for cloud computing, networking and cloud computing, cloud computing technical skills, networking skills training, cloud architect skills, cloud architect training, cloud architect career tips, cloud architect, cloud career tips, cloud career training, cloud as a career, cloud career, aws networking training, classful and classless addressing, what is cidr, subnetting vlsm, vlsm exercises, subnetting, vlsm, go cloud architects

Id: nzXAUufdjes

Length: 222min 36sec (13356 seconds)

Published: Wed Nov 17 2021