Border Gateway Protocol (BGP) and connecting to an Internet eXchange Point (IXP)

I've also put the configuration in for how to create the bird server for the ixp and for the other is at the other ISPs and on that website okay so so my name is dear Miller Breen I'm a lecturer in macaroni university as it's obvious I'm not Ugandan right so I'm Irish and so my background is as an as a CTO and an ISP in Ireland for a number of years and before I came to Uganda and when it came to Uganda I was met by a lovely woman called Dorothea Kelo who is a lecturer in McCreary and she she dragged me out there to work with the net labs research center so the net net labs Research Center is in McCreary University so what I'm going to talk about today I'm gonna look at I was asked by the guys in micro tech if I could look talk about tips about mistakes that could be made and when you're setting up an ISP or in the early stages of an ISP um I was I suppose unfortunate that when I was in my last company we started off with no network and a CTO a CEO and chief financial officer and no network and then we bought 10 companies in about six months and I had the the task of trying to make all those networks all come together so I we bought got all the different types of networks so I can go through what's good what's bad from experience I'm going to look at the inter at anima system routing and it probably would have been better if this presentation I mean before the last one because I'm kind of a going through it at a more simpler level than the the EO IP presentation I'm going to talk about internet exchange points and the benefits of them I'll talk about bgp peering configuration and then finally how to configure the BGP appearing to an IXP so I categorize I was thinking about this when I was asked about well the different types of ISPs so I decided to categorize into five basic types now there are going to be different you know other bits that are in between and a little bit of this and a little bit of that but basically there are five basic am i small ISPs anyway as you become up as you grow to becoming a larger ISP so the first one of them um I call it my kind of NAT ISP this is where somebody has decided that they want to make up our Wireless ISP for example and they they're good with radios they know they can put a radio on roof and they can point it at an access point and they've got a high building in a in a town and they point their radios to that high building and their problem is how do I get connected to the rest of the internet so what they do is they buy themselves a you know a fiber connection or something from a provider and they put on an ass gateway router and basically they are gnashing for all of these customers so the customers are getting private space IPS they're being nasod onto a connection to an upstream provider so for all intents and purposes they have extended the green ISP here they've extended their network inside the other network these guys do not have an ESN they do not have IP addresses they just have private IP which is very very bad because the customers are getting a private IP address in the first place if they're going to have a little router at home then they're going to be netting on netting because they're gnashing on their own router then they're knotting on the ISP so it's bad the ISP finds it extremely difficult to can't multi-home so they can't connect to another ISP for resilience if they ever decide they want to address their network properly and connect to another is speed and they have to go on re IP the whole network so this is kind of I would consider the worst model of an ISP okay the next one isn't much better but it's slightly better it's where I've gone to an upstream provider and I've asked for public IP addresses so there's there they're supplying me a whole cell service they're giving me IP addresses from their blocks but my ISP does not have an ASN it is part of the larger ISPs ASN so they're still only that that ESN they're five three three three and this network the yellow network is the is the ISP we're talking about and they're part of the green network as well okay that the next type one is getting better this is where we decide to be a full ISP ourselves and in Africa they the key here is AfriNIC AfriNIC will assign ASN numbers autonomous system numbers which makes our network autonomous we can then connect to other autonomous systems so we look for and system number and we look for IP blocks ipv4 ipv6 blocks in this case you'll note that the green network and the yellow network are not connected this network now has its own ASN so it's an independent isp it's using this tier 2 ISP up here to get transit ok so this is a much much better model because the day I get kind of cheesed off with the green ISP who are charging me too much I just simply have to make a connection to another ISP and switch my traffic the customers don't even get an outage so this is a good model I can also decide to connect to another ISP anyway and create route routing such that if wouldn't fails the other one stays up so I don't lose any connectivity my next model is what I would consider the best model that's where I also make a connection to an IXP I'm going to talk about ixps in a few minutes so I'm not going to spend long here but basically I have appearing with a transit provider to get to the general internet and I've appearing with the IXP locally to handle local traffic and keep it local and keep latency down so that is my favorite model for an ISP and the last model I would consider the virtual is B and I don't know whether it's that common in Africa but it's very very common in Europe worst brands like Sky Tesco you know shopping centers who want to pretender an ISP they will they will buy the core network pieces of being an ISP and then they will basically buy access from other providers so far like the cuckoo in the nest if you like so if this is say Kenya Telecom or the general ISP they will buy services off them and they will supply them with routers which will establish VPNs generally back to eternal termination around their network so that the customers get IP addresses from the Tesco network or the the sky you network they will appear for all intensive purposes that are connected to this as an independent network the big advantage for the Tesco and Sky is they don't care about installing they outsource that to the to the company you have the real network they don't care about rolling out hardware because that's the other person's problem so they usually then try to layer top of that additional services like TV services so that's very common with TV companies wanting to layer their products so there are my five basic models now you may find that there are some mix of these you may have an isp who's a Wireless ISP who is using fiber network from somebody else so they may have a little bit of the virtual network on their network to you know to get to customers that you don't know why sloughs okay so what is an autonomous system and what is the inter autonomous system routing so here is a kind of a basic picture of two autonomous systems they have an autonomous system number and there are two types of ASN number there are 16-bit and 32-bit and there's nothing really different between them except one is longer than the other and the longer one was brought in because they were running out of 16-bit ASN numbers so inside the autonomous system they can the owner of the autonomous system the ISP is entitled to use whatever interior gateway protocol and they prefer they can use OSPF they can use is is they can use rip if they want to do static routes if they even wanted to and then the interconnect to other providers using ebgp external BGP border gateway protocol and that's the standard protocol being used on the internet since the 1990s since 1994 I've also shown ibgp links and they tend their interior BGP links and they're really used to connect border routers so if I have two border routers connecting to either one other ASN or to our multiple layer sense I can connect my border routers use an eye BGP so that the routes can be shared between those routers using BGP the three kind of main elements to a beach P and packet are the origin what is the origin of the path the AAS path that's a list of all of the autonomous systems that this particular routers pass through so as as a as a BGP message passes from one autonomous system to the other the local ASN is prepended to the list of a SNS so when it gets to the end they will have a full list of all the air sends to this particular route past room and then the next hop what IP address should they should go to next so there's them the router itself has a BGP finite state machine now this is a fairly simplified version of it but it's kind of important particularly for troubleshooting will often see someone would ring and say oh my rouge is showing that it's connected okay there it's in connect state why is it not working well the BGP state means that the state machine must be in one of these states at any given time and what happens when you plug in your cable and if you've configured on both sides is it will quickly jump from an idle state to a connect state onto an open sense they open connect open confirm stage and finally establish but that will happen really quickly it'll go through those and then if there's a fault at any of those states like for example here it's not able to go to open confirm it will go to TCP fails and I will stay in an active state so you may see on your routing BGP peer print you will see that the link is an active state you think okay that's fine now it's actively trying to make a connection so it's not actually fine it's only fine when it gets to an established state so this is just looking at the different messages to go over and back but the main message to to look at here is the update message and as you can see the the things I was talking about the a s path the next hub and there's the NL or I or the network layer reachability information is the network that we're trying to truce so looking at an actual packet wait so this is I've I've obviously ignored the the layer 2 of the layer 3 and the session and I'm really looking here at the application the BGP so in the update message it gives you the attributes for this particular route so the origin is always IGP and the path attribute then if a s path is five nine nine nine that's the a s number for this particular route the next hop is the IP address okay and then we have the network that's been that's been routed for ipv6 it looks very very similar very very similar in fact and difference here is that it'll also show a local scope route so here's the local scope as well as the actual global okay so that's a kind of a quick look at BGP so where does ixps come into the IXP is a really good idea and has been around since about 1992 started in sweden and alot of people think and I heard this in Ireland where either smaller ISPs gone or the ixp that's something for the big boys that's for aircar more for Vodafone or for that it's not for me I'm only I'm too small to connect to the ixb um I would say that's not true and in in my network at home when I told our CEO that I was paying to join I next and it's a lot more expensive than joining and UI XP I don't know the price for joining the Kenyan ixb but we're free at the moment but not for much longer but the INEX was a lot more expensive and you said why are we spending this money and I said well we're gonna route at least 30 percent of our traffic will go off transit which were paying heavily for and we'll go to the ixp and because we have we're lucky in Ireland we have a lot of content providers in country so he didn't believe me but he said look here to CTO you make the decision so I did and he came to me afterwards to say well well as a 22% 23% he was kind of smiling and I said actually it was 42% so we got 42 percent of our traffic that had been gone to transit jumped to the electronics so it was massive for us it was a big big reduction in cost and overtime of course the cost cut up because transit will use going up and open up but relatively we our costs were being kept down because you know ago over a third of our traffic was going to the exchange so this is a makeup of the internet I would say of 10 years ago and there's been a transformation in the internet over the last 10 years and a lot of people haven't really noticed but there has been but this was the original model the idea that you would have access to your ISPs or local ISPs at tier 3 they're the ones that have customers they're the ones whose houses are connected above those we have tier tools these are ISPs that provide services to ISPs and to large enterprise so a bank might connect directly to a tier to a small ISP connects to a tier 2 and then tier ones they're kind of the very large ones that interconnect the continents if you like and then connected to those we had our content providers you had your YouTube you had your Facebook's all of those at the beginning but what problem do you remember from eight years ago eight to ten years ago with YouTube for example I'm a big monster ropey fan I live in Uganda I can't watch monster rugby on O the odd time I can watch on DSTV now but generally speaking I can't watch it in DSTV so I have to pretend I didn't see the score wait until Monday night get onto YouTube watch somebody has put the match up put it 10 years ago I would have seen this circular thing going on all the time buffering buffering buffering buffering you don't see that so much anymore and that's because the this has changed so what was the problem well the problem was that generally speaking if we had these consumers here and one of them was trying to connect to this large enterprise let's say in Uganda or in Kenya so somebody in in in here in Nairobi was trying to connect to Standard Chartered Bank and both of those companies have their services here in Kenya what was very likely happening was that the traffic would go to the local ISP who went to a tier 2 is B who happened not to be the tier 2 ISP supplying this large enterprise so their traffic headed out beyond the sea at Mombasa into the Middle East true to Europe into links in London and then got switched at links in London and came all the way back so your traffic was going from Kenya to London and from London back so you can see if I told you that I wanted you to go from here down to into the middle of town you wouldn't go via London right you'd go in along that lovely big highway they have there outside outside on the road yeah so why you would you put your packets that way we're getting latency there's cost because the transit has to be paid for and the growth of the ecosystem within Kenya then it doesn't have that same own if you have you had your traffic local so what the ixp does is it allows the local providers to connect together on a switch it started out as a switch with two providers telia and telly - there were the two providers in Sweden connected to a single switch provided by Sun net and in the UK the second exchange in London with five providers connected to that exchange so it started very small and it simply was a switch and everyone peered at everyone else and so now my traffic comes to the rousers here and they make the decision that actually is not efficient to send it to to Europe it's much more efficient to go to the IXP and then down to the large enterprise and in a lot of cases the large enterprise will actually have their own connection now as an IXP ixps are no longer the preserve of ISPs its ISPs large enterprise banks etc anyone who is an ASN so we reduced our latency reduce costs and we increase our autonomy in the region so what the ixp does today and the IXP has too many members to just have a switch and everyone peering at everyone because every time you appear with everyone you have a lot of peers to set up so it's much more efficient if the ISP provides a route server and a root server is a bit like a ruler it takes in routes and a route comes in from for this is p4 is p1 into the exchange and the the root server here sends that rules on to all the other ISPs but unlike a Rooter it doesn't prepend it's a yes to das pad so if it was a regular router it would say ok what is a ma am I in I am in a SN three seven one two three I'll prepend back to the path which is a s two four three four so when it gets to the one at the end it should see in the a s path it should see and three seven one two three and then it should see two four three four but when it's a root server in exchange it doesn't prevent this so for all intents and purposes this is P receives the root as if it got it directly from this one but it doesn't have to appear with this one it just peers with the root server okay so looking a configuration so very start off we're going to look at configuring is p1 ASN five one one one and we're going to do a peering configuration with this tier 2 ISP called is p3 we're not going to look at the configuration at this but we're going to look as we are the ISP connecting to an upstream provider now the configuration for this is in those files that I talked about on the website so if you wanted to rebuild this testbed or something it's fairly easy to do so we put some IP addresses on so we have IP address here for this client representing all the customers and we have one 99.1 that 1 to 102 a 99 : 1 : 1 : : 100 so there's our IP on the machine representing the customers of this is P we obviously have dot 1 and Colin : 1 of that particular subnet or that network as the interface IP here and we have one 99 dot v dot v dot 11 and 2 a 99 : v : v : : 11 as the IPS on the network true to here and I've done it nice and simply I just used slash twenty fours and slash 64's I know because the point-to-point link we should be using slash 30 or whatever but just for simplicity the other thing I've done is I've created a loopback interface here and I've created a loopback interface here and I'll explain why I've done that when we get through to the configuration so the first thing anyway I'm configuring ipv4 and ipv6 so the ipv6 package is not enabled by default on the on the micro ticks so to enable that I run the command system package print it shows me all the packages and I see an X beside IP IP v6 so I simply do system package enable to because it's package number 2 and it enables it ok I then default the box so on system reset configuration no defaults is it equals yes I don't want the mikrotik defaults which just confuses my configuration later on so I clear out all of those and it reboots and the first thing I do want particularly in a test bit is I give the the boxer name so system identity set name is p1 so when I'm working on it later the the prompt and the shell shows me the name so it's kind of easy to work with the next thing I do is I create a bridge and I call it loopback 0 if you if you if any you were to our Cisco trains you'll always be your loopback is there by default mikrotik it's not the loop back is very very handy so why would you have a loop back so if you take a router and it has got five routes into the router and I use the IP address of one of those routes to represent my connection to that router and should that physical interface go down not alone as that interface gone down my link to that router has gone down the loopback interface doesn't have any physical interfaces it's just routed as a as a kind of a floating interface on the router so if any one of the interfaces goes down as long as I have a pathway true one of the existing interfaces I can get true to the loopback but further to that we use router IDs for routing protocols OSPF uses a BGP user so we have to have a routing router ID so when you're running a testbed and if you root your loopback interfaces and if your router ID on the instances for routing are the same as your loopback interfaces it's extremely easy to figure out where routes came in the router table so that's why I do that yeah so the loopback interface I raise it by creating a bridge and just calling it loopback zero you can call it whatever you like but loopback zero makes sense so interface bridge add name equals loopback zero I didn't give us an IP address as per the diagram I just had their IP address at address 200 that one that one that one and it's typically a slash 32 because I just it's a network it's a host network rather than actually having a full network there's no need we only need one IP address on it because it doesn't have any interfaces and I give it the interface loopback zero so creating this so that becomes a name for you know associated with this bridge I didn't put my IP addresses on the interfaces again as per the diagram so Ethernet to is getting their IP address 199 dot one dot one dot one and the same interface is getting this ipv6 address to a 99 : 1 : 1 : : 1 I do for Ethernet 5 are the networks that were connected to the green network that the isp 3 so I can look at my addresses I've configured IP address print and I see my addresses and if I do ipv6 address print I can see all my addresses so now I have IP addresses on the system but have no routing so have no routes so what I must do now is create an instance a BGP instance and I give it some sort of a name you can call it anything Mickey Mouse or whatever you like but it makes sense to give it a name associated with the ASN number so I've called it a SN 5 1 1 1 and the AS IS 5 1 1 1 that was assigned to me by afrinic and then the router ID is the IP address of the loopback interface on that same router it can again it can be anything once it's unique but again it's just much much easier for troubleshooting if it matches the loopback interface on that router so now I've created the instance so the instance is running but it's still not sharing any information so I now almost held the instance what should you share and I want you to share the networks 199 dot one dot one dot zero and two a 99 : 1 : 1 : : zero there are the two networks on my client side if you like my customers networks and I want to share them with the upstream provider so if I do a routing BGP instance print I can see all of that stuff that I've just put in there so all of them the stuff of just literally put in here and if I do routing bgp network print I can see the networks that I'm routing so I'm routing the two networks that I described here okay so we're routing to the other side but we're accepting kind of anything back as well so if the if the ISP 3 is sending us routes we're kind of accepting everything that's not sensible there's a lot of routes we don't need to receive back and thankfully there's an RFC six nine eight zero that defines a lot of networks that we should not be receiving ok so our special purpose IP registries and are listed nicely in that RFC we can go pull down the RFC and see what they are so we're going to build a filter for those but the other things we should reject we should reject receiving our own prefix why would we receive our own prefix through our interface so from outside I should not be receiving your own prefixes it's not clear enough the other thing I shouldn't be receiving is default routes we're doing full routing with ISP upstream so why would I want to receive default routes so I want to discard default routes and then I'm going to discard all of the networks that are defined in 6 9 8 0 and that's the minimum or I should do I'll talk about something else is in a second but we should definitely do those things those things will serve to protect us in some way so I create a filter my first filter this these two lines here these are as you can see I'm creating a chain I'm calling it in ISP tree - ipv4 again I just like logical names for 4 things you could call this Mickey Mouse again if you like it doesn't matter what you call it and I'm saying for the prefix $1.99 there's one that's 1.0 which is my own prefix discard so anything that comes from isp tree to my router that's for my own network inside i said well actually no actually right we shouldn't be receiving routes for a network i know about already because it's my own network so we discard and we do the same for default rooms who send any routes that we're receiving that are for default we discard them so that's the two lines in the in that piece of the filter and then I wanted to create a new filter called RFC six nine eight zero and don't worry about all of these they're very very easy you just go to RFC six nine eight zero you pull it down it gives you all these addresses so you know it's not some fancy stuff I made up this is stuff that's in the RFC so we create in exactly the same way we create a filter put it's got a new name it's called in - RFC six nine eight zero - ipv4 and then when I apply this filter I'm going to put a line in here it it says once I've done this filter jump to this filter so I've kind of made it like one big long filter I could have made it all actually one big long filter buts just harder to to look at it's easier to look at that knowing what six nine eight zero and what are my own routes and my default route so I've created the filter but it's not doing any filtering s and I'll show you that in a second so the second set of filters I'm doing is exactly the same process for ipv6 again I go to six nine eight zero it gives me all the ipv6 prefixes I shouldn't be receiving incoming I say I don't want to receive my own network I don't want to see default routes and when you're finished doing this better jump to this better okay so now I need to apply them so to apply them I say routing filter add the chain is the one I just defined oh no that's wrong I should slide ahead sorry so routing BGP peer ad and is p3 so this is up here I'm adding for the route this is in is p1 adding it for our peer is p3 their tier two provider the instance is my own instance name which is a SN 5 1 1 1 and the remote is is 5 3 3 3 so I'm adding appearing for this is p3 that's the tier to the remote address is the IP address of the router in is p3 and then I say my in filter is in - is p3 - ipv4 which if you remember was those three lines at the top and then it will automatically pull in all those other lines 6 9 8 0 because it jumps to them once that filter is applied I also create an out filter I'll look at that out filter yeah John passed it I'll show it in on a second I'll do exactly the same process for ipv6 and the only difference for ipv6 is you see you must specify address - families equals ipv6 apart from that is to save except obviously I'm using the ipv6 addresses and the ipv6 filters I created an out filter and it's probably not necessary but again I like to do things tightly so the out filter I'm saying is only route for one 99.1 that 1.0 invert match equals yes so that's saying discard everything that's this that's not this this is the only thing I want to send out so I'm discarding everything but the inverted match which is my own network and why am i doing that I don't want anything inside my network somehow advertising something I don't want to advertise opt is p3 so I'm filtering only my network to advertise upstream and the same thing for ipv6 I'm only advertising that as well so I do that and hey presto I have a route so I can see routing BGP peer print and now I can see I can see that five one nine nine dot v dot v dot tree which is the other side the remote is is 5/3 tree tree so it's there I can see that I have a BGP route learned now I'm learning one 99.3 that's 3.0 which was the client name IPS 4 is p3 and the Gateway is 199 dot v dot v dot 33 okay and it's using the default distance so now very quickly I get routes learnt in from that ISP in reality if I was connecting to an upstream provider I'd be getting in lots of routes because I would learn all the routes they know about as well so my testbed obviously a money learning one the other thing you can do is and there is a team Comrie this is correctly pronounced Comrie most people don't know that this is actually the Welsh for Wales Wales is a little country hanging off the United Kingdom on the on the west of the United Kingdom between Bru England and Ireland and the Welsh name for Wales is Comrie but it's spelled cymru and there's a project team called team Comrie and they create what's called a polygon list and what the bogan list is is a list of what he called martians they're continuously looking at the rme they've the privada the ripe AfriNIC etc finding networks that are legitimate but networks that have not yet been assigned so if afrinic have networks that haven't yet assigned why would you be receiving them as routes from is Petri you don't want to receive them as routes Maya's Petri so they kindly provide a list of all of those but rather you happen to configure them yourself all you have to do is actually BGP peer with their servers so you can BGP be here with either 6 5 3 3 3 at port 888 or at 6 5 3 3 2 at port 888 and automatically they will blacklist all of those networks that you shouldn't receive they call them merchants as in networks that are coming from Mars and we don't have any networks on Mars yet and it gives you a full configuration for configuration that peering there at the team Comrie website so I'm not going to steal their thunder okay so we've configured now our transit so we're happy but what about the ixb we want to connect to the IXP locally so we're now looking at connecting we have is p1 connected to is Petri and now we want to connect I p12 the exchange the internet exchange and what we should start seeing is roots coming from this ASN which is also connected to the exchange so looking at the IP addresses so the exchanges bird server bird is the Internet routing daemon that's used by most IXPs and it's it's addresses one 99.9 dat 9.1 and 2 8 9 9 : 9 : 9 : column 1 and what I've actually done is I have created an LXE container and in that I have this here is actually in LXE container that can install onto a Linux host I have that LXE container up on the website so if you wanted to actually pull it down and set this up you can I've also created an OVA so that you can do it in VirtualBox if you like as well so that's the exchange setup so the exchange now I have to configure each and at one with these addresses and have to create a bgp peering true to one 99.9 that 9.1 at ASN 5 9 9 9 so simple enough IP address I add my addresses 199 at 9 at 11 and 2 a 9 : 9 : 9 : : 11 on each one so now I have IP addresses I should be able to ping 199 at 9 at 9.1 and to a 9 9 : 9 : 9 : : 1 which is the internet exchange so I create filters again and thankfully I've already created the 6 9 800 filter so I don't have to recreate it again I can just jump to it and the same here and the filter here I'm basically telling the link going towards the I XP I don't want to receive these were either my own network or default routes from the ixp ok it's exactly the same logic as I explained earlier on and I don't have to show you the six nine eight zero filter because it already exists we used it already for for is p3 so I create my routing Pierce or routing PGP p-- here I'm adding the name I'm using ixp the ASN is my own one so I'm the instance is a SN 5 1 1 1 is my own BGP instance and the remote AAS is 5 999 the remote address 1 9 9 dat 9 dat 9.1 which is the bird serve running on the exchange and my filter now is in ixp - ipv4 and if you remember that jumps to the six nine eight zero filter so we get all of that filtering and we do exactly the same for for ipv6 and we apply the same out filters for ipv4 and ipv6 again we don't want to advertise something we don't actually know about so now when we look at a root table now we can see that we've learned two rows 199 - two - that's zero we do not have a BGP connection with us and ona SN v 2 - 2 are the ISP - we have no connection we've only got the connection that it is also peers with the exchange so we're learning from the exchange about that particular rows as well so that's pretty much it I just wanted to go through the pitfalls you might have with starting an ISP and I urge you if you're starting an ISP pay the $300 - AfriNIC get your a SN get your IP block and your ipv6 block don't take IP addresses from another provider because if you do you're stuck with them it's a small bit of effort to set it up first as you can see it's not very very large amount of effort to set up those pairings ok so set yourself up properly with a transit provider and once you are happy and you're running with one transit provider connect to another transit provider have to do resilience so if you have a problem one has a problem you're still on you're not going trying to tell your customers that it's not my fault it's a fairy comes fault connect to your local I XP keep your traffic local another thing that I XPS have come to do in recent years is whole caches so you'll find Google's GGC cache Akamai's cache and all of these caches been kept at the exchange so you get the benefit of all of that downloads that have been done by other people in other ISPs are now sitting at the cache so you're very very low latency from your customer to the cache ok so that's another that's a really good reason to join the ixb apart from letting your customers access eat government services which will all be local and other local services as well keeping content local reduces latency reduces cost and yeah you'll find everything at that address or WWE bring calm for such Microtech you'll find out all of the stuff I've just I've just talked about okay any questions yes sir thank you very much why do ever sorry I didn't get that Oh Mickey Mouse and I don't know maybe when I was growing up on his kid and I like to watch Mickey Mouse running around and Italian black-and-white you know nowadays you get him in color and stuff but I know I like Mickey Mouse a nice dad didn't like Tom okay thank you very much we have another presentation now I also wanted to remind you that at any moment when you like you can go outside right downstairs we have coffee and some snacks but the next presenter now is Bosco Mulva and he will be talking about my critic in universities [Applause]

Channel: MikroTik

Views: 4,895

Keywords: mikrotik, routerboard, routeros, latvia

Id: WgEolc05_hg

Length: 38min 1sec (2281 seconds)

Published: Mon Feb 05 2018