ENCOR - Cloud vs On-Prem Deployments

happy wednesday everybody welcome back to the encore study group we are moving into blueprint section 1.3 today which has to do with the uh well the on-prem versus cloud deployments is how it sells it out in the blueprint we're going to be exploring exactly what that means for a lot of us network engineers we don't have a lot of data center experience and so it's interesting that cisco tossed this into the encore blueprint but they did it with a reason and that is because we as network engineers if we want to be you know complete network engineers able to solve any problem we need to understand what's going on inside the data center so that when our organization has problems around the networking that's in there that we can help with that and more and more organizations are going to have not only their data center to worry about but also a cloud deployment so again it is mission critical that we as network engineers understand what's going on number one in the data center what's going on in the cloud and kind of how it all bridges together so stick with us today we're going to have a lot of fun unpacking this concept of cloud versus on-prem deployments and looking at again as network engineers what we need to know about that so here's our agenda today we are going to be looking at oops where did it go there it goes we're going to be looking first and foremost at what a virtual workload is so again i'm going to approach this from the perspective that you know you may not have a whole lot of experience around virtual workloads and virtual machines and such if you do have a lot of experience with that uh hang with us because we're not going to spend a lot of time on that just want to lay the foundation before we get into everything about cloud so from there we're going to talk about on-prem data centers and everything that goes into an on-premise data center and then that's when it starts to get good right because then we dive into the cloud what is the cloud and for a lot of us you know if we are asked to define what exactly the cloud is might not have a good answer for that and so we're going to give you a solid definition of what a cloud service is so if you want to know what the cloud is hang with us we're going to solidly define that leave no doubt in your mind um from there we're going to talk about kind of this talk about the world of cloud we toss a lot of cloud terms around a lot of those cloud terms are going to fit into either a cloud service model and you know spoiler alert that would be things like infrastructure as a service platform as a service we hear these terms i as paths and sas tossed around that would be a cloud service model and then we also have cloud deployments cloud deployment models such as public and private and hybrid clouds so now if we have time i'm going to go into a little bit more detail on what the difference is between the different cloud deployment model types but you know i mean we'll just see how we do here we've got an hour to unpack as much of this as we can and again as we can see we've got a full slate as usual so let's go ahead and dive right in oh and just a reminder as well this video has been pre-recorded so even though it's being premiered live on youtube i am in the chat so if you have any questions at all about this section of the blueprint or really any section of the blueprint again this is a study group so if you have any questions about spanning tree or eigrp or anything like that feel free to toss those questions into the comment section into the chat and i'll do my best to respond to that and answer that but of course also if you have any questions as i'm explaining things here on the whiteboard chalkboard i suppose um you know feel free to chime in with those into the chat as well all right so moving right into the world of cloud which really starts again with the world of the data center so let's just go ahead and define for us what exactly a virtual world looks like because it's kind of of virtualization is not unique to the data center it's not unique to servers there's network virtualization the concept of virtualization is taking a piece of hardware and carving it up into sort of virtual instances of that hardware and there should be a very popular network virtualization technology that we are all familiar with and if you're not familiar with this you probably shouldn't be studying for the encore you should probably be looking at the oh i guess the c cent is gone but the ccna or the even the cct this concept of a virtual local area network or a vlan all right we're most of us again if you're studying for the encore we are all aware of what a vlan is a vlan is just a broadcast domain right it's it's a subset you know we could talk about it being a subnet broadcast domain all these different definitions but really what we're doing is we're taking a physical switch and carving it up into multiple virtual switches so to speak a physical switch has a cam table what does every vlan have has a cam table it's like it's its own little private virtual switch and you know as a port comes in we assign it to one of those virtual switches so to speak so and i'm using the term virtual switch that's not the right word for it it's just an example but um we're bringing ports into a virtual local area network so we had a physical land now we have a virtual land the exact same concept is going to apply in the world of servers we used to have this concept that we would need a single physical server for most applications in our network and the story i usually like to tell with this is you know i i've got this pc running right now i've got my drawing application running i've got my chrome running over here with all my notes in it i've got over here my my obs running that's doing the recording of course if i look down in my task bar i've got so many other applications running right now i probably should have closed some of those before i started recording because applications suck the resources on a pc and so even though pc operating systems like windows and even mac os and linux i didn't mean anything by saying even mac os by the way and just you know example anyways any pc operating system is going to be able to handle running many applications at once and server operating systems are the same way if we don't have experience with server operating systems guess what we're going to find a lot of the same uh the same examples we have windows server windows server is simply windows for the server world we also have linux and well really that's that's the primary two operating systems we're gonna see out there there isn't really a mac os in the server world but either way we can run multiple applications if this is my hardware let me um let me do this i'm gonna erase this real quick give me a little bit more room to draw this out so the idea wait pen working good all right so the idea here is that if we draw this like a stack i've got a piece of hardware that would be actual metal an actual server or an actual pc or what have you and i'm going to install an operating system onto that and i know that again my pc right now is running a lot of different applications application here application here again chrome obs all of these different apps are running all at the same time the operating system is built to handle that but in the enterprise space this doesn't fly why doesn't this fly well the the reason it doesn't fly is because if i install multiple enterprise applications like my email server onto an operating system and then maybe my a web server and then maybe a enterprise like an erp system or something like that if i deploy all these different enterprise applications onto one server into one operating system instance number one you're going to have some call software or soft problems in that they're all going to want cpu they're all going to want memory and there might not be enough resources to go around and you're going to call you know this application vendor up on the phone and say hey your application's running slow and they're going to look at your operating or look at your server and say wait you're running all these other applications call us back when our application is the only application running because we can't control those other applications and we can't control the amount of resources that they need so that's kind of like the soft issues here's a hard issue we only have so many network ports to go around and if i install two different applications that are both running web servers even if it's just you're going to access the application via a a web tab i pull up a new tab in chrome and i you know go to use this application via its web interface well all of them need port 80 open potentially or whatever port i'm going to open in that range and and they can't all have port 80 if port 80 lands on the server it's got to know which application to give it to and only one of those applications is going to uh be applicable for that or be available to that and so we start to actually have some hard limitations about which applications could actually run on the same operating system instance so this is a problem we can't run multiple operating systems in the server world and so what we really end up doing is if i just erase this again a little bit what we really end up with here is this very whoops there we go this type of stack one application to one operating system instance to one piece of hardware and this was the data center world in the mid 2000s every single application had its own operating system instance and its own piece of hardware and this just was a terrible use of resources we'd have pieces of hardware that are running just a web server and so it'd be sitting at like 10 cpu cycles and maybe like 15 percent of its memory and so we're buying all the cpu and memory and it's just sitting there wasted not to mention hard drive space by the way but that doesn't really tie into this conversation either way the world of virtualization came into play just again sort of like think about this from a networking perspective if i needed multiple broadcast domains back in the early 90s what was i going to do i was going to have to buy three separate pieces three separate switches let's say for three separate broadcast domains so i could have hr and finance and it or what have you on three different sets of switches and and i might buy three different 24 port switches and i only really need like five ports on each switch so i deploy those three different switches i've got my users one two three four five on each one of these even though these are all 24 ports so i'm wasting 19 ports on every single switch wouldn't it be better to put all 15 of those ports on one switch and then use vlans to segment the broadcast domains okay we understand this from a networking perspective the same concept applies to virtual machines in the virtual server world so this is what we're going to do we're going to instead of calling this a piece of hardware we're going to maintain the relationship but okay we still need one operating system instance to one application that ratio needs to stay the same but instead what we're going to do let me just clear this make it easy on us all right so instead what we're going to do is we're going to and take one piece of hardware at some point we live in a virtual world but at some point we have to deploy hardware okay it just is the reality we're going to take one piece of hardware we're going to install on it instead of an operating system we're gonna we're gonna deploy something we call a hypervisor now for you and me who aren't you know we're network engineers we're not server people you can think of the hypervisor as the operating system that's basically what it is you know you get a new piece of a new server a new piece of hardware and i can install windows i can install linux or i could install a hypervisor hypervisor think of it again as an operating system trust me you know we would dive into the weeds server people and virtualization wizards would have probably a lot of issues with me calling it an operating system but again just at a high level that's basically what we're doing is we're installing this hypervisor and you've probably heard of some of these hypervisors very popular ones vmware vmware makes a product called esxi most of us just call it vmware we don't call it esxi but if you've heard either of these phrases this is what this is it's a hypervisor microsoft makes one microsoft's version is called hyper v hyper dash v i don't know why they called it hyper-v i mean clearly they took like the first six letters of hypervisor and just called it hyper-v i guess i have no idea but um kind of an interesting one there and then red hat by the way has you know they they take linux which is an open you know uh framework i suppose an open source and there's an open source version out there called kvm well they have an enterprise version of kvm kvm would be an open source hypervisor so it stands for kernel virtual machine the word kernel with a k is a big linux concept so kernel virtual machine or kvm these are probably the three biggest ones citrix does have one out there called zen server but they're not major players in the space um at least any i'd say anymore but i don't think they ever really were usually if you're going to find things that are supporting different virtual hypervisors it's going to be one of those three and so from here what we're going to do is now we're going to install multiple instances of operating systems lots of different instances of operating systems on this one hypervisor this hypervisor is able to handle the installation of of many different instances and these instances are called virtual machines or vms again a lot of us even if you're not super familiar with any of these concepts you've probably heard of this concept of virtual machines and there goes my camera for those who uh for those who are regular attenders of this podcast or podcast study group um you're well accustomed to the fact that my camera can only stand for 30 minutes at a time so we're well i must have turned my camera on early because we're only less than 15 minutes into this video but you can rest assured that it will be turning it back on again in 30 minutes so um my operating systems are virtual machines that's basically what it is a virtual machine is an operating system instance and they have to be hosted on a platform that can support many different operating systems at once and that would be that hypervisor and then again remember that one-to-one relationship we're going to install our applications onto each one of these operating systems now if you're paying attention you're looking at this and saying okay jeff that's great and all you just told me we had problems from an application proceeding multiple applications sharing the same piece of hardware because of resource contentions and that is true but um and so some of those problems could still absolutely remain but the reality is in most cases once we took those applications and split them across to those many different pieces of hardware remember what i said about like you know cpu instances might be running at 10 and memory might be running at 15 percent well if we load this piece of hardware up with a ton of resources when i say a ton of resources i'm meaning like you know modern day servers are going to have like 32 cores or more of cpu processing 256 gig is probably the minimum amount of ram that we're going to deploy regularly it's 384 512 and even a terabyte of memory is installed into modern day virtual hosts and so we have a lot of resources and we're going to put as many applications as many virtual machines really at this point as many virtual machines onto one piece of hardware as we can from a resource contention perspective and the hypervisor usually has tools to help us manage the resourcing so it's going to know that hey we're peeking out our memory we're over 80 memory that's not great you know maybe we want to consider moving some virtual machines around which we can do okay so that's problem number one problem number two you're saying okay well jeff wait a second that's the soft requirement what about the hard issue with like ports and we can have multiple ports right i mean if i've got you know port 80 open on this operating system it knows to hand it to that application and i could do the same thing here and that's well and good other than what about the actual network connection the physical network connection that's coming in and we have a destination port of 80. well what are we going to do with that well we're going to switch that like normal but the way we're going to do that let me erase the word hypervisor here we're going to do that is we're going to have now a virtual switch or a v switch this virtual switch is truly a virtual instance of a switch that virtually connects all of these different virtual machines and so when this physical network connection comes in it's going to get mapped to a port on that virtual switch and you know even though we've got destination port we also have a destination ip address so that's well and good what but the switch is running at layer two well again we're gonna have a destination mac address as well sorry i just looked over and realized i wrote right underneath myself i always i always forget to cross that thing off okay so what i'd written earlier was red hat kvm that kernel virtual machine all right very good so um so that destination mac address is going to tell this virtual switch which has an actual cam table and it's going to have an actual you know mac to virtual port mapping it's going to see that destination mac address and forward it to the appropriate virtual machine so even as multiple ports or multiple connections arrive that are destined for the same port well all of those packets are going to be destined for a specific mac address and they're going to be forwarded out the appropriate interface at that point these virtual switches again are usually layer 2 only they're not going to be doing layer 3 switching svi work all of that is still going to be upstream of the server which is you know backwards in my drawing because i just wrote a down arrow but upstream of the server is where we're going to have on the physical switch infrastructure is where the svis are going to live okay so that is the world of virtualization if you have any questions at all let me know um so this is a very zoomed in view of a virtual host let's zoom this out a little bit okay next agenda item here we're going to be talking about on-prem data centers so before the advent of cloud and even in this day and age of you know the world of cloud we have this concept of an on-premise data center a data center is really going to store two of our most mission critical assets even you know more and more so becoming the most true most new critical assets a company can own number one is its data that is something that didn't use to show up on uh financial reports but nowadays it actually shows up on some organization's balance sheets their data has a tremendous amount of value a tremendous amount of worth that could be company ip that could be customer information um that could be you know yeah like by ip i guess i mean things like their designs their patents all of the company trade secrets whatever is accessed on a regular basis by its uh staffing okay so whatever our data is is stored usually on hard drives in the data center and the other thing too would be our applications so our applications is less less of something that's going to be of external value but it is the wheels by which our machine our our company churns so if for example our data center were to go down yeah nobody has access to the data but what about the applications can a nurse check the emr system for information on a patient no they can't um can receptionists check people in a lot of cases no can sales people access their again their their um their customer their crm tools no i mean if we don't have access to our applications we can't do business on a day-to-day basis and you know that's even excluding communications applications you know yeah email we all still use email but things like slack and cisco spark or webex teams and microsoft teams and all these communications uh applications are also down now you know not to mention sharing applications a little sharepoint such the list could go on and on clearly these are two very mission critical assets again for every organization and so the reason why we store them in the data center is to keep them online and available at all times that's the goal of a data center and this is why we see data centers often securely locked down sometimes even with guards posted at a lot of big data centers not everybody has access to the data center in most cases it's behind the you know lock and key or scanning your badge or what have you and there's usually some pretty good um siloing as far as hey the server people can go in and do this and the networking people can go in and do that and you're not to mess with each other's stuff okay we deploy networks with massive amounts of redundancy so i guess that's the purpose of what i want to talk about here is everything that goes into managing a data center so everything we just we just talked about security that would be physical security sure but also network and cyber security so making sure our servers are all running antivirus and making sure our firewalls are configured and you know we have to worry about east-west security in the networking world that would mean can two servers on the same subnet on the same vlan talk to one another and do we want them talking to each other and can we stop them from talking to one another i mean that was a that's a huge topic alone and that's just the first thing we're mentioning here i just started to mention that the network is going to be highly redundant that is absolutely the case we want to make sure that we have extraordinarily resilient switches in there cisco makes nexus switches for the data center nexus to this day has a more resilient operating system than ios the way they architected nx os is very different than ios it's why they run two different operating systems they feel mostly the same there are some differences for sure but ultimately nexus are just designed to be better switches more resilient switches plus they're going to have protocols that are run only inside the data center like fibre channel fibre channel over ethernet and virtual port channeling are all going to show up in the network we of course are going to have servers now we have to be careful with the word server because we just looked at the fact that we have these virtual machines which are virtual servers we also had physical servers so what we're going to find here is that the physical servers we're going to call virtual hosts i believe i had already said that once or mentioned that term but a virtual host would be a physical server that hosts many virtual machines and the virtual servers are either going to be called vms but there is technically a phrase that we use called guests and most as far as vernacular is concerned i don't i hear all the time virtual hosts be said i don't hear a lot of people refer to virtual machines as guests but i have i mean and i do and others can and that's a perfectly fine phrase so whether you're here it's called virtual machines or guests either way that would be uh the two different kinds of servers to manage in this environment so on top of the virtualization of the servers we have to worry about that hypervisor so the hypervisor whether it's vmware or hyper-v and whether we want to migrate from one to another and keeping it all up to date the software and the patching we large organizations will sometimes just differentiate from the server team and the virtualization team and that's because they have way more on their plate with just managing the virtualization the hypervisors they can't actually help manage the servers they just don't have time for that oh storage don't forget storage so i mentioned earlier that hard drive space is an issue well more and more in modern networks or modern data centers we don't see hard drives directly attached to servers instead we'll have a lot of servers that are all accessing through the network a storage device so this network storage appliance we think of vendors like netapp emc dell has a whole line of server or storage devices a network-based storage array is going to put all of the hard drives here and none of the hard drives up here and so it's kind of weird to think about a pc or a server booting up without any hard drives attached they technically can there's ways that they can access a boot drive and boot directly off of a network attached hard drive this takes that concept like we mentioned earlier of having like maybe 10 percent 50 45 of of their hard drive space consumed i mean if i have 10 terabytes on this one server and it's 45 consumed then i've got 550 gig that's just sitting there unavailable and maybe over here on the left instead of 10 it's at 99 full and i could really use some of those gigabytes over here and so the way we do that is by deploying network-based arrays another concept here would be hyper-converged infrastructures for those who have heard of cisco hyperflex that's a a big platform that cisco offers that basically provides this what's weird about hyper-convergence is everything i just said is sort of undone because we moved the hard drives back into the servers but yeah i got to be careful here i'm going to go down rabbit trail but the hard drives are not managed by the servers the servers don't think they have hard drives attached instead we have special software that more or less borrows the server shells to say i'm going to store my hard drives in these servers but i'm going to turn around and present the hard drives back to them as a giant pool of storage so if that made sense great if it didn't then trust me it's a much longer conversation and worth having and go watch some youtube videos on hyper convergence especially cisco hyperflex which in my opinion is is like catching lightning i mean it's a phenomenal phenomenal product way better than any of the other hyper-converged platforms out there but i might be a little biased who knows um the on top of all that we've gotta this kind of ties into security a little bit um just from the perspective of we're trying to keep everything online um so i just think about like yeah anyways it has to do with this part right trying to keep everything online uh backups and disaster recovery two very distinct concepts that often get grouped together but again mutually exclusive backups are all about restoring lost data disaster recovery is all about getting back online as fast as possible in the event of a disaster we can use backup systems as our dr plan it's not usually the best idea but uh you know again it's it's all about keeping us online not losing any data i mean it's bad enough that you know losing a spreadsheet you've been working on for an hour right i mean that feels like a sock in the gut the last thing we want to do is lose months of data or weeks of data or even days of data especially when you know things like ransomware exist out there and it could lock us out we have to go really far back in time to recover data i mean that's bad news we want to keep all of these systems in place as best as we can in order to keep our again right here our data and our applications online and accessible okay if if this conversation didn't result in your feeling like your head is spinning then uh maybe data center is a great place for you i got to say i love the data center i i mean i have this conversation i could have all day long i love data center technology i just i've spent a lot of time doing data center designs and data center deployments and again talk about awesome technologies like cisco ucs which is a server platform or hyperflex which is hyperconverged platform i love netapp storage anyways it doesn't matter what my personal vendor preferences are all that to say it's just it's really fun to bring all of these different technologies together and make them work but there is a certain amount of stress involved because if we lose data if we our applications go down our organization suffers and i've spent some time talking in previous study groups about this idea that we need nit we get so wrapped up in the bits and bytes and technical i don't know technology i guess we get so wrapped up in our own stuff that we forget that we are maybe a medical clinic that services patients or a financial clinic a financial clinic a financial organization that helps people with their retirement accounts or um maybe we're a restaurant chain and we actually serve food to people i mean we we have this tendency to forget that we have a front lines you know we're we're really the guys in the back you know the man or the woman behind the curtain so to speak that's who we are and and our job is to keep that online because if this goes down our customers hurt our our organization hurts if we lose enough money people might lose their jobs i mean this affects real people and that's where the stress can come in so all that to say oh yeah about halfway good perfect we're on schedule here all that to say in a lot of cases this is why organizations are looking at cloud solutions because cloud solutions say all right all of this right here this is a headache and this is stressful and i don't want to manage this myself and so companies like amazon is a very popular cloud service oops sorry amazon is a very popular cloud service provider microsoft very popular cloud service provider a lot of different organizations have said hey move your virtual machines up into the cloud and at that point we will do all of the other things so if we look at this if if i just move a virtual machine into a cloud space then i don't have to worry about security as much they're going to propose present a lot of security technologies to me in other words like i might not have the capability of managing east west security inside my own data center but in the cloud i do because they've got the systems in place to make that happen and all i need to do is make a few clicks and boom my security profile is deployed the networking i don't need to worry about network switches and performing network upgrades and scaling out to more buying more network switches to support more servers the servers themselves i have to manage hardware anymore dealing with hardware failures and procuring new equipment what if by start to run out of server space well that's not in the cloud space that's not for me to worry about that's for the vendor to worry about all i know is my virtual machines are online and functional i don't care how many physical servers they have i don't have to manage hypervisor anymore because again i'm just managing virtual machines and same thing with storage so pretty much all of this goes away to some extent in a cloud service now that said as you can imagine um you know i like to liken this concept to a valet service you know i mean i've only used i think i used a valet service once um but either way it's it's it's nice it's a convenience right you you pull up to a hotel and there's somebody standing outside and they're like hey for ten dollars i'll take your car and i'll park it for you and then when you're ready for it i'll go get it for you and bring it back to you and so you don't have to walk all the way to the parking lot or what have you and you can unload all of your luggage and suitcase and everything right there um those valet services are really convenient i don't have to park my car i don't have to retrieve my car i can unload everything right there but it comes at a cost i have to pay 10 bucks um i can self park for free but i can pay somebody else to do a bunch of stuff for me okay cloud world is the same concept there is a big misconception out there that cloud is cheaper somehow that i move all my virtual machines to the cloud and because i don't have to buy servers i don't to buy networking and i don't have to buy hypervisor licensing and i don't have to buy all these systems that somehow it's cheaper well no it's not cheaper i'm paying for all that stuff you know the cloud service provider is installing network switches and they're installing physical servers and somebody's paying for that and guess who's paying for it okay this guy you and me we are paying for it and so any clouds cost analysis over the course of five years or even over the course of three years in most cases you're going to pay more money for a cloud service but look at all of the things that you don't have to manage anymore that is the value and that is why sometimes it's worth it to move into a cloud space okay there's also this concept of capex versus opex spend we don't have time to go into that but the gist of that is cloud also is just kind of that monthly subscription i pay maybe fifteen thousand dollars a month regardless of anything whereas in it i might not have to pay nearly that much a month of up until i get the call you know i'm worth the tech right so we have to make the call to the cfo hey cfo i need another 100 000 worth of network switches or servers or what have you and so you know their their life is dealing with this kind of flow where i'm spending very little and then boom i spend a bunch of money in one month and i'm spending very little and then boom i spend a bunch of money in one month those are capex spends and cfos as you can imagine don't like living with that stress any more than you or i would managing our personal finances i'd really rather subscribe to a service for 10 bucks a month then once a year have to pay 100 a month yeah i end up spending more i spend 120 in the year but from a finances perspective it's a whole lot easier to manage a ten dollar monthly line item than it is to manage once a year boom 100 bucks like that's that's hard to to budget for that hopefully that makes sense i said there wasn't enough time to cover it and yet i covered it so there we go okay moving on we are ready to talk about do do what in the world a cloud is okay i promised we were going to offer a very clear definition and we are in fact we're going to borrow somebody else's definition and that would be nist and nist is an american organization that um for uh ah i'm blanking on the uh that it's the national institute of it's not science and technology standards and technology that's it they like to define standards within the world of technology and so even though even though it's a national um i'm sorry even though it's an american institution these they created a definition of terms of what the cloud is that is used pretty much worldwide okay it's a pretty good set of definitions it comes down to five different concepts so what is a cloud service a cloud service is going to have these five characteristics according to nist okay first of all it's going to have this concept of on-demand self-service on-demand self-service what is on-demand self-service so here's the concept each one of these i want to compare it to a traditional network you know versus what it modern day is let's say we have an application developer and a traditional network and that application developer needs a new virtual machine okay do we usually give our application developers and our um i don't know our hr reps and our financial team and our legal team do we give them direct access to our technology interfaces and the answer is typically no so how is the poor developer supposed to get a virtual machine well traditional networking in a lot of cases they'll call the help desk say hey i need a virtual machine the help desk will say all right cool i'll enter a ticket now we've got that ticket in we will get your virtual machine and then somewhere between maybe an hour and maybe three days later they get their virtual machine now i don't know about you when i call up i mean yeah think ahead and do your due diligence and do your planning that sounds good but i tell you what there are times when i'm like alright i'm sitting down to do my work and oh no i didn't i didn't call that person and tell them i need this um all right hey um hey can i get that virtual machine spun up like in the next hour oh it's going to be three days okay well shoot and you know now so i'm all ready to work on it even if it's an hour let alone you know a three day period right like either way it kills my momentum and it kills my productivity to have to sit and wait for that service to be delivered to me okay compare this to a cloud service a cloud service imagine if amazon who spins up virtual machines for people imagine if i had to call amazon and say hey amazon i need a new virtual machine imagine how many calls they'd be fielding imagine how obnoxious that would be for most people no instead we know we get this i'm going to pull up a website and i'm going to click a button that gives me a virtual machine that is self-service what does on-demand mean on-demand means anytime i want middle of the night weekend whenever holidays christmas morning i want to spin up a virtual machine amazon service is there spinning up virtual machines on christmas morning god bless them but it's but it's automated systems that are doing it um i was i was speaking to a um just a local municipality once about their cloud experience they were using a local cloud provider and one of their big frustrations was yeah anytime we want to add spinning virtual machines was on demand they could do that but anytime they wanted to like let's say add storage to their storage pool they had to call the company up and that company ran eight to five monday through friday that was it so saturday afternoon they're going in to do a cut over they're going to do this big upgrade and they find out they don't have enough storage like now now their upgrade is canceled on a saturday night because they can't for they can't provision more resources into their cloud pool this wasn't an on-demand service right it was it was it was self-service yes um but it wasn't on demand they could yeah yeah i believe that is how it worked i had to catch myself there they'd log into the system and they'd provision you know request more more resources but somebody on the other end had to actually accept it so all that to say they were looking at moving away from that provider because it wasn't convenient enough you know again in amazon or even meant even most regional or local cloud players will be on demand these days okay so on-demand self-service number one number two broad network access broad network access basically means i can access it from anywhere now the most common example here from a traditional network perspective would be vpns okay we in the networking world we're very comfortable with the concept of a vpn right i'm going to fire up my vpn and connect to my business network and and then i've got access to all of my resources users are not super great with vpns it's just one more process that they have to do one more application and again they don't understand why they don't understand what a vpn is like most of us in the networking world we understand what a vpn is why why it is and how we access the vpns so it's hard for us to you know we have that cursive knowledge concept where it's hard for us to step into the role of a user who doesn't understand those things and so they're needing to vpn in to the network in order to access something so let's say in our private data center we had on-demand self-service and we can do this in a private data center um we cisco's got tools that we can deploy that allow people to go out and provision their own virtual machines it's fantastic but what if i have to use a vpn to log in to this on-demand self-service well that's no longer in this definition that's no longer a cloud service so one thing to keep in mind in all this is if we want to call our data center a private cloud like a lot of places do a private cloud needs to have all of these characteristics otherwise it's not a private cloud it's just a data center okay so on-demand self-service broad network access again accessible from anywhere i can i can access amazon or microsoft from the starbucks from my house from my company network vpn in vpn out anywhere i can access these resources so that is broad network access incidentally by the way that is not to say that my virtual machines are accessible anywhere i should not be able to access my virtual machines from anywhere in the world that's a big security threat it's saying that the service is available from anywhere in the world okay resource pooling come on there we go resource pooling so resource pooling is more or less what we've been talking about with saying that i just let me think of an example here so so the idea is saying i have let's say 100 terabytes of storage okay i don't care where that 100 terabytes of storage list it could be 15 terabytes on that storage array 25 terabytes on that storage array 10 terabytes on that server i don't care i just know i have 100 terabytes of storage and every time i install a new virtual machine or every time my virtual machine goes out and downloads a bunch of data that resource pool takes down same concept with cpu cycles and memory consumption and even network bandwidth in a lot of cases the idea is simply here's your pool of resources use it how you see fit and and this is a challenge in traditional infrastructures because you know okay i'm sorry here's why because there's one more aspect to the resource pooling has to do with that story i told earlier let's say i have 100 terabytes and now i'm approaching 95 96 terabytes so i'm just going to go out and i'm going to grab another 50 terabytes and put it into my infrastructure so i can just increase the size of my pool and that company that municipality they were not able to do that seamlessly well in a in a private cloud so to speak in a data center solution it's very hard to do this because we might be able to pool our resources as best we can but there's no guarantee that we're just going to be sitting on more resources so i might have a brand new storage array it's 100 terabytes and provisioning it out and life is good or life seems good and then my hr department goes out and buys a new application that needs 50 terabytes on its own and so um i might have to go out and buy more storage or support that application that's it's really hard to get away from that because oops told you it's going to go away it's really hard for us to get away from that in the data center space because we don't want to just sit on a bunch of hardware we don't want to go out and buy you know an exabyte of storage just in case we need it but that's what all the cloud providers are doing they're sitting on exabytes of data so if you need to go out and spin up an extra few terabytes or whatever i mean they've got that sitting and ready and available at any time and so that's the advantage of going with a public cloud for example all right the other aspect by the way and this is sort of a sub bullet point is this concept of multi-tenancy all right multi-tenancy is well at least the idea of multi-tenancy is that i've got a lot of different clients running on one physical infrastructure so as you can imagine again amazon and microsoft they're going to have one physical server with virtual machines from my company and virtual machines from another company and virtual machines maybe from a direct competitor right of mine they don't care they're just throwing virtual machines out onto servers which means by the way that we're all storing data on the same physical hard drive which means we're all using the same physical network uplinks multi-tenancy means we're all sharing resources in this environment but we have the security of not being able to access each other's stuff okay so that's the concept of again resource pooling multi-tenancy is kind of a part of that conversation okay rapid elasticity i think i jumped the gun a little bit on the resource pulling concept obviously these two are tied together a little bit but the idea wow this is a heck of a y what did i do there that's amazing elasticity all right there we go so rapid elasticity more or less means i can stretch my environment up and i can compress it down at any time so if i need 100 terabytes today i well you know a better example might actually be um if i need let's say one 10 terabytes of memory i'm using 10 terabytes of memory across all my virtual machines today tomorrow busy season hits it's christmas season or prom season or whatever it is that makes my business spike up and all of a sudden i need uh 15 terabytes 50 percent more for this season so i'm gonna pay which we're going to get into next by the way so i'm going to pay for this 10 terabytes and then i need to stretch really quickly to 15 terabytes for this season for three months let's say and after three months boom i go back down to 10 terabytes this elasticity up and down is a key part of a cloud infrastructure so what about rapid elasticity that's what we that's what we mean by that okay and the last component this is fancy it's called measured whoops measured service sounds fancy basically means i pay for what i use okay at home here i i pay for my water use and i pay for my electricity and i pay for maybe gas consumption natural gas and i pay for what i use if i use less electricity if i were to actually shut down my studio for a month i would probably use less electricity that month and i would pay less okay same thing with cloud they need to be able to measure how much you spend or i'm sorry how much you consume so that they can bill you for the appropriate amount i don't want to have to pay for twice as much electricity if i didn't use it so if the electric company is has bad measuring practices and they think i'm using more than what i spend then i'm going to eventually well i don't know if i can leave my electric company because there's kind of only one in town but you get you get the point right with the cloud world i can just flip my servers over to somebody else and so they need to have accurate measuring tools in place so that they can appropriately bill appropriate bill appropriately bill according to what i use okay in the private data center space this is like it's an interesting one okay there is a use for this in in on-prem data centers and the concept is basically this i might have different departments let's say hr and finance and it's another department legal team i guess and what we can do is this concept of um chargeback yeah let me put it up here i don't have much space i gotta not draw underneath beneath me today okay um charge back this concept of chargeback is saying that i'm gonna track what each department uses in my in my data center space and i'm going to charge back to those departments what they consumed now that sounds really weird to say i'm going to charge a department for consuming i.t resources but the interesting thing with this is that it is typically its own cost center and i.t is generally viewed as a drag on the bottom line you know i'm as an organization i've got to pay my i.t staff and i've got to buy it equipment and it just it's just a drag on the bottom line and what those companies a lot of companies tend to forget is yeah yeah you spent a million dollars in i.t stuff but that million dollars went to enable your hr department it went to enable your legal department and want to enable whatever and so this concept of chargeback basically what organizations are doing with it is they're removing it as a cost center and saying it costs zero dollars and it spends zero dollars it's just it's neutral which should be actually part of our mindset anyways but it's not often because of how much it costs in the world of i.t so it's no longer a cost center instead i'm going to make it budgets as part of each one of these departments and so hr now has an i.t budget spend and so let's say we i gave them of my million dollars i gave them 200k so now now hr department has 200k well i'm detecting how much you've spent and you actually spent 210 000 where's that extra 10 000 coming from well it's coming from your cost center hr and now you can't buy as nice of a christmas party for your team i guess i don't know uh christmas bonuses go down whatever the situation is right but it shouldn't be my fault if i'm in i.t and we're all used to this right we need another network switch oh we need another server oh we need another extra set of storage is it usually it's fault i mean every now and again we don't plan ahead very well but in a lot of cases the growth is the growth and if the hr team deployed a new application we weren't expecting and we need more servers as a result well that looks really bad on me to go in and say uh mr cfo i forgot or i didn't realize it was going to happen and i need more servers um i know we just bought fifty thousand dollars in servers but i'm getting another thirty thousand in servers based on what just happened and i can try to say it's because of hr in most cases it's not one thing that caused the problem or caused the cascade it's a little bit of this a little bit of that a little bit of another thing maybe five different departments all added just enough so that now i've got to go buy more stuff that i wasn't expecting and again that capex spent concept the cfo wasn't expecting it and they're not going to be happy about it and i look like the bad guy in this chargeback model we eliminate all of that okay hr spent a little bit too much financial finance spent a little too much legal spent a little bit too much and the good news is you know if they all spent that 210 000 instead of 200 well now i've got a 30 000 budget that didn't show up a thirty thousand dollar budget that i can go out and buy new equipment that i got ten thousand dollars from each department's cost center so i can go out and make the necessary spends and now if anybody's the bad guy it's the people who use the resources but generally speaking as much as anything it shows an organization where their technology spend is going and the value that it brings so that's why i'm i'm a huge fan of this i've not seen it deployed a whole lot outside of very large enterprises but it is it's a really neat concept to be able to say it is no longer a cost center it's no longer draining on the bottom line it is now just simply a part of every department's budget and chargeback is how we get there now the downside to chargeback is it's really only applicable here at least easily measurable from a cloud perspective it's hard to deploy this or push this out to all um all technology stacks you know pcs and network switches closet switches etc but they can be done they it definitely can be done okay coming up on our time here so let's move on to the next topic we're going to kind of blaze through these might go a little few minutes over we'll see shouldn't be too bad i did not add a new layer there we go okay so cloud service models we're going to talk about yeah let's just talk about cloud service models okay this is a pretty important concept because it gets tossed around a lot okay um not all clouds are created equal if a cloud service has those five characteristics that we just saw on the previous screen then it's a cloud it's a cloud service but not all cloud services do the exact same thing so that's where we get into this concept of cloud service models and cloud deployment models and usually we can categorize a cloud service as you know into various options here cloud services models mentioned them earlier infrastructure as a service sometimes the a's are capitalized sometimes they're not we have platform as a service and we have software as a service the gist of these service models is going to be how much do i want to manage myself um i don't know how to very quickly me i don't know i don't know if that worked or not what do i want to do that's what i'm trying to say here what do i want to do infrastructure as a service is interesting because i have to manage we talk about virtual machines a lot all i have to manage is the virtual machine that seems pretty good all of that stuff on the first slide the the networking the servers the storage i don't have to manage any of that i just have to manage a virtual machine and that sounds good except from a cloud perspective that's actually the most i can manage they're not going to let me manage network switches and firewalls and storage arrays and such on their network that makes sense okay but can i manage less that's the question i'm going to skip platform as a service for a moment because it's a little bit nebulous a little cloud humor there um hard to define platform as a service but you you kind of get a sense of what platform as a service is once we define software as a service so what software is a service software as a service says i'm only going to manage the software itself so when we think about it what does a virtual machine look like well virtual machine has an operating system and it has an application so that's interesting so what am i managing from a infrastructure service perspective ding ding ding i've got to manage the operating system do i want to manage a bunch of windows server instances do i want to manage a bunch of linux operating system instances that's the question i have to ask what do again what do i want to do what do i want to manage okay infrastructure service says i'm going to manage the entire virtual machine operating system application deployment i've got to install the application myself etc software as a service says i don't have to manage the operating system anymore i only have to manage the application cool that's actually pretty neat now a great example of software as a service the reality is that many of us use software as a service at home on in some level um software service usually has the mark of not having to install anything onto your machine so i know that about a year and a half ago my wife and i were done managing budgets on spreadsheets we decided to sign up for an online budget or well call it online or trying to avoid the word online failure uh a budget software program now used to be ten years ago you'd go out to the store and you'd buy a budget application and you'd take it home and you'd install it onto your machine and you'd do all of your budgeting on the software instance but nowadays i'm using a software as a service that means that i didn't have to install anything i didn't have to buy anything i just subscribe to a service and i log in via the web and i do all my budgeting on there all of the cpu and all of the memory consumption everything that that application takes is in somebody else's data center it's not on my pc and plus i can access my budget from my pc i can manage it from our family pc i can manage it from my phone because it's just it's software as a service the software is running somewhere else i just have to manage my own budget right so that software is a service so infrastructure service software as a service we see the difference here where does platform as a service ride because like there's i mean there's just an application operating system instance i mean it looks like it's right between the two and it is it's sort of between the two what we have to realize about platform as a service and something i didn't mention earlier us as network engineers might not fully appreciate this not every application requires only one virtual machine in a lot of cases i might need three different virtual machines to run a particular application one of these virtual machines this one might be a web server i i did it again didn't i oh well we'll just cross this one out and we'll put it up here application operating system so one of these services w-e-b one of those services servers one of those virtual machines might be a web server one of them might be the data server what we call the application server it's actually running the application and then one of them might be a back-end database and so a platform as the services is usually saying i'm going to take part of an application like the database and i'm going to manage that for you and then i can take that database concept i can apply it to many different applications so it's sort of in between in that like i'm still managing some operating system instances i'm still managing the application itself but i'm still managing less so it's still less than infrastructure as a service i mean this is the most that i'm managing and this is the least and platform as a service is usually somewhere in between okay do some google searching to look at different examples of platform as a services i think that's about all the time we have for today last but not least are the cloud deployment models unfortunately we do not have time to go over these in detail but as always i mean this resource is all over the place not the least of which is in the cpt nuggets course on encore that will give you all the information that you need to know all right cloud deployment models what are these these are these concepts that we've already mentioned public cloud private cloud hybrid cloud and yes there is a fourth it is called a community cloud just to run through these real quick a public cloud simply means that it is somebody else's it belongs to somebody else so this would be amazon this would be microsoft there are literally hundreds of cloud service providers out there right pick one they my virtual machine lives inside of somebody else's data center that's what a public cloud is or you know given what we just talked about it's not even necessarily a virtual machine it's a service a database service or a software as a service that is a public cloud instance private cloud means it's inside my data center the key thing that i always stress here big deal is that just because i have a data center does not mean i have a private cloud a lot of people refer to their own data centers as a private cloud that is not the case unless it's got those five nist characteristics do we have on-demand self-service do we have measured service do we have all of these five characteristics and in i dare say most cases we do not okay um so so if we don't have at least the majority of those characteristics we can't call our data center private cloud or we can but it's not a private cloud so a lot of organizations will put a lot of effort into making their data center a private cloud allowing their developers to log into a web portal and spin up virtual machines and such i mean that's that's all really good practice um hybrid cloud probably what it exactly what it sounds like um and that warrants a little bit more time i'll come back to hybrid cloud in a moment community cloud um i might get in trouble for saying this but it's unless i don't know in most cases you're not going to run into community clouds this is almost something you just need to know for exams okay where you might actually see community clouds in action would be in the higher ed space so colleges and universities uh you might see community clouds you might occasionally see them with medical research but you're not going to see a lot of community clouds out there in the world at least certainly you're not going to be interfacing with a lot of them unless i guess you already have one but the idea of a community cloud is simply me and a set of organizations we're all sharing resources i guess i should have defined that we're all sharing a data center space it's just the three of us or five of us or what have you um and we're gonna share these resources that we're all paying into or what have you okay so this idea of hybrid cloud the only thing i want to clarify with this is yes it means we have public cloud access and in some cases it means we have multiple public clouds so maybe we have amazon and microsoft spun up and then we also have a private cloud and so we've got all three of these we've got virtual machines stored on premise we've got virtual machines stored in amazon we've got virtual machines stored in microsoft and that is more and more the reality cisco calls this the multi-cloud world multi-cloud is sort of an expansion on hybrid cloud but their point is simply saying that we've got more than one public cloud vendor at play and usually what we're going to try to do is deploy a i know it's barely on there deploy a brokering piece of software that runs our data center um yeah anyways and it's going to help us decide where to spin up a new workload so i got a new application i got to spin it up where should i deploy it should i deploy in my private cloud should i deploy in my public cloud where is it going to be the cheapest to run and if it's going to blow the resources out of the water in the private cloud or it's something that's going to spike regularly it probably doesn't belong on premise but if it's a very stable resource load and it doesn't take up a whole lot it's probably gonna be cheapest to deploy into my private cloud and so i simply leverage the cloud that's most appropriate now i just described that from a cost perspective every pi every public cloud usually has specific reasons why you'd want to use it i mean that's why you maybe want to use amazon and microsoft they both have features that the other one doesn't have and so there's always going to be reasons even outside of cost as to why you might want to use one private cloud or one public cloud over the other so um that that is that i i thank you everyone for for coming and for engaging on this topic it's a it's not it feels like it's not a networking topic um because it's very data center servers uh server-minded conversation and topic but the reality is is that even though i didn't get to really go into the details of this if i've got a private cloud and a public cloud i have to have networking going between them a lot of times it involves mpls circuits being terminated into public cloud or a vpn and so we as network engineers yes need to understand the networking but before we can even understand the networking we need to understand why we have multiple clouds what even is a cloud and that is this conversation that's exactly in my opinion why cisco has included this on the encore blueprint because if you're going to be a ccnp you need to understand what the cloud is because your organization probably has access to the cloud so again thank you very much for joining next week we are doing the oh this is exciting drum roll please the principles of sd-wan a lot of people have been asking are we going to cover sd-wan as part of this study group and the answer is yes we will august 26th that's in two weeks from this live showing be sure to tune into that because it's going to be the start of a conversation around cisco's sd-wan product and it's going to be a lot of fun there's a lot to cover with sd-wan as well so everyone i hope you have a great rest of your day and we will see you next time [Music] bye-bye [Music] [Music] [Music] [Music] [Music] [Music] [Music] [Music] [Music] [Music] you

Channel: KishSquared

Views: 1,073

Rating: 4.9111109 out of 5

Keywords:

Id: CuaVDhLkgls

Length: 70min 50sec (4250 seconds)

Published: Wed Aug 12 2020