MikroTik, BGP and Internet Exchanges

I think everybody's kind of filtered in so my name is Justin Wilson this presentation is going to be kind of a bgp deep dive over the past year or so I've been talking to folks who I look at there be BGP configs and like well you could tweak this you could tweak that and they're like well I never never knew about that so I wanted to kind of go over some some common things you can do to influence your your BGP sessions we're going to talk about some mikrotik specific stuff we're going to talk about mikrotik and IX is I'm going to assume by this presentation you know a little something about BGP but if you don't know big deal I always try to tailor my presentations so you take away something you know maybe in you know three months four months six months whatever when you're you're actually getting into doing this if you're not doing it now you can say oh I remember when Justin talked about in that presentation I'll go back and look at it on YouTube or look at it on you know the mikrotik side or whatever so I got a couple links for for folks who want to learn more about BGP so I'm not going to bore you with some of the the basics of BGP so Who am I pretty much all the links to everything I'm involved in you can go to j2s w.com hi I redid it between whispah and now it's just been a couple in a couple weeks I've been in the ISP industry since about 1993 I'm a managing partner of fibre data exchange FDI X we got a little table out there so if you want to want to talk about peering and I exes feel free to come by I've started a podcast past several months it's usually three to five minutes a week on things ISPs need to know I just do an audio podcast I post that on the the MTI em blog anybody who was here earlier for the brothers whis I'm fairly active in that when I when I have time I'm a huge GI Joe collector I've had clients pay me in GI Joe's before so I'm fine with that and I got some certifications you know whatever to show you I'm such a hardcore geek this is my dog here he's a 13 year old Rottweiler his name's router that was that was the first name that came to mind so let's let's let's talk about filters let's talk about BGP filters here so what you're talking about BGP there's two ways to look at your traffic you know how the internet views you and how you view the internet so you have your outbound traffic how you're viewing the internet and how the rest of the internet views you most of the tricks we're going to talk about today or how you can influence your outbound traffic how others see you and how you can influence that is a little complex you can do it but it gets into a lot a lot of details a lot of knowing what your provider can do knowing what their provider can do it gets into a lot about the internet when you're when you're troubleshooting when you're seeing how the internet views you and how other providers you view you there are things out there called looking glasses looking glasses are your friend when you're trying to determine you know hey how does my traffic transit through Hurricane electric to get to cogent to get to level three so you know just like just like our two little guys arguing over here you know is it six or nine it just depends on your viewpoint so let's talk about filter rules how many how many people here are running BGP currently okay good good number so filter rules can make or break your BGP stuff mikrotik rules can be buggy anybody who's had lots of filter rules in their mikrotik routers know that you can get some unpredictable behavior at times one of the tricks I always do is I'll drag around a filter for a couple times and kind of kick it in the button it's just I I don't know what it is it's something that a lot of people complain about but what we're going to talk about today will help mitigate that a little so you know what's going on on your filter rules so if you have other Network admins they know what's going on so one of the takeaways I want you to do for today is if you're running filter chains look at consolidating what you have now your prefixes your your access lists all of that consolidate them into their own chain I see people do jump rules I see people do prefix lists perp here that sort of thing we're going to work on change your thinking a little so in the same thing for the the peers that you accept prefixes from consolidation is the key so so what are we talking about here let's let's talk about how I like to set up a filter rule so we have a we have a peer here Hurricane electric what I typically do is we know there a SN so when we create our filter rule as6 9:39 that's hurricane electric so that's what hurricane electrics in this case they're out chain is going to be then we do a friendly name because you know some of us can memorize a SN numbers more than we can remember you know our anniversaries for our wives you know my wife was just in here so I can I can say that then we we give it a friendly name hurricane or however you want to know know it Hurricane electric HED whatever so then we want to know is it ipv4 or is an ipv6 filter I like to break out ipv4 and ipv6 into their own chains and we'll we'll talk about that in a little bit and is it an in and in or an out chain so in my hurricane electric here I have an in chain and I have an out chain and I have one specific for them and anybody who's ever had me work on their routers or seen my configs on the blog or anything I'm a fan of uppercase letters it's not that big of a deal in comments and things like that but when you get to where you're duplicating things or you're moving them over to a router having either all lowercase or uppercase is a nice thing to have you don't have to worry about miss typing things into me uppercase just looks a little a little better it looks cleaner but you know that could be me so why are we you know why are we doing something simple like that you can say well you know I have my own way of doing filters which is fine as long as they kind of meet this criteria you know so as you scale your filters need to be manageable I have customers that have 30-some peers on a router with 300 and some filter rules in in that particular router so you can imagine when you when you go looking through stuff even-even if you sort them by the the chain it gets a little you're I start getting crossed a little the less typing you want to do on these chains the better because that eliminates not only you know router overhead but it eliminates human error you know the more typing we do the more mistakes we're going to make I'm a big fan of includes instead of jumps so I've seen some some folks do filter chains where they get to the OP the end of the chain they they jump to another chain fall through that chain and then jump back into the main chain that gets a little messy too if you don't take anything away else from this slide take away the make sure you have explicit denies on your outbound stuff there's nothing like triggering a prefix limit or triggering you know something that your your provider shuts you off and you have to give them a call and you know troubleshoot with them so on your outbound the the thing to do is you specify what you're doing and then you deny everything else you know it's like when you lie to somebody you know and you use you your specific then you just I don't know so on your inbound deny it's it's kind of the opposite you you deny everything that and we'll get to this in a little bit you deny everything then you allow whatever is left over so it's a specific deny inbound and then you just allow everything else so here's here's one of the the chains I was talking about this is one of one of my clients that I work with pretty regularly this is their own IP space so we have one filter chain that we apply to every peer that they have so when they add some IP space or they want to hey I don't want to advertise these specific blocks or whatever they go into one place they edit it right there and it gets pushed out to all their peers and and some of this we're going to bring all this together so I just want you to kind of you know soak this in and we're gonna bring this all together and a little bit so when you have inbound traffic you everybody in here should be dropping you know 192 10.17 - all this stuff from your outbound peers so you want a sanity check chain that says you know I'm gonna filter out bogans I'm gonna filter out all my 192 s I'm gonna filter out all the private IP space that sort of thing you also want to filter out your own IP space you don't want someone to advertise your own IP space back to you or you don't want to you know somehow if you have remote peers or if you have multiple exit points to your network you don't want to advertise your own IP space back to yourself the Internet when we get to looking at prefixes just about any backbone provider these days they don't advertise anything smaller than a 24 so if if you know from cogent or hurricane electric you're getting something smaller than a 24 something's wrong so we want to drop anything smaller than a 24 on our inbound now if you have a special set up with a you know neighboring ISP or something like that and you're doing BGP back and forth between yourselves maybe there's a use case for accepting something smaller than a 24 but usually not so one of the the things that we we always run into with BGP is how do we make our route table smaller the global routing table is growing each and every day we have ISPs buying their own 24s from the the third parties so in there they're injecting those into the routing table and they can't really summarize those routes because the the last time I bought a 24 for a client it was 21 dollars an IP so they paid a little bit over five grand for that block well if you can only do that once every couple years you're probably not going to get blocks that you can summarize so one of the tricks that you can do and and I like to do this on some of our really busy routers is I will have my upstream provider I will have them send me a default route and I'll have them send me a full routing table and then what I will do on the router is I will filter out anything smaller than say a 23 now that kind of defeats a little bit of the purpose of getting a full route table but it saves your with with mikrotik and a little birdie kind of showed me that maybe this is changing we'll see we'll see here pretty shortly you know stay tuned to the the brothers wispier but with mikrotik bgp there's this convergence time the more and more providers you add on when something changes a provider drops or you bring up a new peer there's what's called convergence time on mikrotik that's been a sticking point for quite a while now it takes 15 20 sometimes even longer for BGP to converge well what happens when BGP is converging if if you don't have a default route you're probably dropping traffic so one of the tricks you can do is hey I'm gonna get a default route from my provider and I'm gonna filter out anything that's smaller than a 23 so what happens is your your routing table decreases so now your convergence time decreases but if it's if it's smaller than a 23 how do you anybody know how it gets out to the internet default route yep so when we bring in when I when I talked about earlier when when we were talking about chains vs. jumps you know when we go back this is a downstream provider from this particular company Divi networks they're they're an exhibitor here we accept probably 40 some prefixes from them with this one route statement and one chain i can advertise all 40 of their prefixes to all 30 of these peers and if they say hey we have a new prefix we sold this prefix we did whatever with this prefix I go in one place and I get rid of it so you know anybody who's ever seen any of my presentations or talked to me knows I'm very kind of philosophical about network design there's lots of ways to do stuff so so why are we doing some of this well we want easy to maintain prefix list we want to be able to maintain things pretty easily you know change once and it affects multiple peers you know like I said earlier less typing equals less mistakes you know as your as your network grows and you want it to scale you also have to hit have to think about the hit by a bus you know what is what if my network I got hit by a bus what if I got if I'm the network guy what if I get hit by a bus I want someone to come in and be able to see what I've been doing and have it be consistent have it be something that they can edit pretty easily so now let's let's let's take a deep dive I've given you guys kind of an overview so now let's let's piece some of this together so we all know here's you know here's a BGP with with one peer we just have a connection to a cogent a hurricane electric somebody like that you know here's a here's a to peer setup and we'll get into this a little a little later here's a here's a complex setup we got a couple providers they may have a provider and we're hooked up with an internet exchange so first of all how does BGP determine which route it's going to choose and we're not going to go through a deep thing into this because you can you can read about this all day long on the internet but we're going to kind of do the cliffnotes version so the first test is is the next hop can I reach it you know is it is it alive so then you have what they call synchronization if synchronization is enabled the router will ignore I bgp routes which are not synced and so the third is Cisco specific Cisco uses a weight attribute and with this the largest weight wins after that if the local if the weights are the same the highest local preferences is chosen from local pref and local preface something where we're going to kind of do a deep dive into here in a minute or so next the router checks to see if any of the possible routes were originated locally so it says hey can I can I get to this locally next if two or more routes are equal the router looks at the AES path now I'll think of a s path is kind of a when you do a trace route the trace route does the the router interfaces that you're going to the a s path is the autonomous system numbers or the networks that you're going through to get to your destination so the shortest a s path wins and the router the router knows this you might not see this in the routing table but the router the router knows how how many AAS paths to get to the destination next up the origin attribute you know if our path links are the same BGP selects IGP over external BGP and it's and if it's an incomplete route you know that's that's the very last one BGP then looks at med values med values are something where we're going to another thing we're going to talk about here in a minute the thing to remember here is the lowest value wins so med is where if you have two paths to equal paths to the same destination the med path can influence that then BGP will we've we've already beforehand we've already said hey prefer some IGP routes if it's getting all the way down here it will say okay we've already exhausted the IGP routes let's prefer the external routes over this we only look at the external routes because we've we've already exhausted if it's an internal route so then our last last three things I GP costs your internal BGP are compared to the next hop routers the the closest one wins after that it looks how old is this route the idea here is older routes are more stable I'm gonna trust the older guy you know he's the sage on the hill I'm gonna trust him if if all else fails there's a router ID what the heck if you if you have a lower router ID I'll pick you you know that's the last resort so you know I just wonder you understand why these are important because when we start looking at the meds and the local crafts you can kind of see okay how does this fall through and what what influences what so rightfully so in in router OS the actions and BGP actions are where the action happens of those of you doing BGP how many of you have a 32-bit ASN do you know yeah it's a and it probably been issued in the past four or five years 32-bit bit ASNs are treated a little bit differently so since some of you have some 32-bit ASNs will will kind of we'll talk about those in a little bit so BGP actions this is the focus of what we're really going to talk about and most of our commonly tweaked items are here so the first one is BGP wait if we remember in our past selection BGP wait is a cisco proprietary thing mikrotik is included it because hey if we if we want to talk to Cisco which is a big player on the Internet we want to talk to them completely so BGP weight is not exchanged to other routers it's only between you and the router that you're paired with it doesn't go beyond them the thing to remember is the highest weight wins so the higher number wins so I have a cheat sheet on my desk with a lot of these because some of them the higher weight wins some of them the lower weight wins or the lower number wins and this this has to do with your outbound path you know where am i sending this traffic and what what a lot of people use this for is failover so if you have you know two ISPs and say they're both 1 1 gig ISPs and your networks only doing you know 500 Meg's or so and you're like hey I like this is P I got favorable terms with them maybe my per bein with cost is is fixed with one and another one they charge me on burstable so I just want to use the other one for for burstable so on our outbound we can we can do a way if they're a Cisco shop we can we can use the weight to influence our outbound traffic next this local press again in our scale down this is something that influences your outbound traffic so the default is a hundred in this the higher number wins so here's what I do with my local press strategy and again remember you know remember here the highest number wins so if we start at the bottom and kind of work up so if you assign a local prep of 400 to your local routes it's going to choose them first hey hi if it's local to me choose it because it's its local next is our customer route so if you're appearing with downstream customers they're they're buying IP from you or you're in a data center or something like that where you're just peered together on a private cross connect a P&I you can say I'm going to prefer routes to my customers next and then if you're on an IX you can say okay after I've filtered through my local routes my customer routes I'm gonna prefer the IX routes next because usually this is a direct pier with you know someone over an IX fabric and then finally we choose our transit routes so our cogent sar our hurricane electrics are level threes and so the thing to remember here is on local prep the higher number wins now if you want to get granular and you this this usually happens a lot on our transit routes say you have cogent you have hurricane electric you have you know four or five providers and you're say hey when I send traffic out I want to perfer cogent first I want to prefer Hurricane electric second you know for for whatever reason maybe maybe you found over the course of time that you know one's a little bit were more reliable than the other weather where you're picking them up maybe it's a little a little more reliable whatever so you can say ok and again remembering the highest number wins or yep highest highest number wins so if I want if I have 4 providers I can say if I want everything to go out hurricane electric I'll make their local pref 104 the next one I can make 103 102 101 you know whatever it doesn't really matter it's not like those who are familiar with OSPF where you have to weigh the cost it's it's absolute that the highest number wins so pre pins pre pins used to be a big deal on the internet what a pre pin does is it artificially adds an AAS path into your when you advertise into BGP the last several years or so a lot of companies are stripping bgp pre pins off so if you if you read an old cisco CCNA book or CCNP book they they talk about how pre pins can influence traffic Kevin Myers from IP architects did a talk at whispah know a couple years ago on the details of BGP pre pins and the research he's done where level three strip some other ones ignore it other ones when they see a prepend they may assign a real low priority to it there they're not a good way of influencing traffic these days so when you when you start talking pre pins and you you Google you know how do i influence my bgp traffic pre pins come up quite a bit just know that they're not that big of a deal these these days it the idea behind them was you you add those AES hops into it so it makes it less desirable so bgp meds we kind of touched on this a little earlier so when when it's making the routing decision when the local prep and the AAS path link are the same for two or more routes toward a certain prefix so if we go back here to our example here so say we have we're trying to get the network c and we we can go to it via provider a or provider B so BGP med is how we can influence which provider to send the traffic out maybe provider a is the cheaper provider and you want to send as much traffic out provider a that you can the provider B or maybe provider a doesn't have a congested port to our network C like our provider B does maybe provider B is 20,000 user ISP and network C Netflix or somebody like that and they're sending all this Netflix traffic to them wellyou've when you you know through your customer complaints and stuff you notice that whenever Netflix traffic goes over provider B it's real slow and people complain but if you switch it to provider a then things things get better and so that's where where meds can can come into play you're you're telling you're telling the the upstream provider hey when when I have to consider two or more routes to you know the same ASN or neighboring ASN choose this way choose this med here the lowest value wins so if you guys notice it gets a little confusing you know sometimes the lowest value wins on some of these things sometimes the highest value wins so that's why I have a little cheat sheet on my mom my desk okay local pref highest you know med you know that that sort of thing so let's let's talk about communities communities are a way to even kind of group all this stuff together so communities have have are supported by almost any provider worth their salt so you being an ISP it's not as important as your upstream supporting communities because they're the ones you want to influence traffic you want to send signals to them you want to you want to send them stuff that you're not sending 600 filters to them you're grouping all these filters into a community and this is where you can send communities to your upstream and it can influence how the rest of the internet sees you remember in earlier we're talking about there's two ways there's how you view the Internet and how the internet views you well communities are one way you can manipulate how the internet views you and you can you can automate communities quite a bit if you've ever heard of a black hole community you're under a DDoS attack you can send a special community to a lot of upstreams that say I'm advertising the IP that's getting attacked I'm going to advertise it via a community to your black hole server and then the provider deals with it from there instead of it coming across your your backbone links so in our earlier slide this is the same same setup as ours our transit peer customer routes but now we've applied that to a community and usually the way a community works is we have an ASN number and then the value so so in this in this case we're sending these communities to our upstream - or our upstream router - to affect our local press so now we can create a community we can group all of our you know our transit routes our peer routes under that community and just send one community to our to our peers any any questions at the moment be glad to I think we're doing ok on time so I'll be glad to answer any questions before we get into IX stuff here this will be the last last section of the slide and got questions feel free to ask so how many of you are familiar with what a somebody over here yes do you do any IRR filtering based upon what the carrier sent to you yes so so on networks and up streams that support it we will do routing registry filtering and for those of you who don't know a routing registry is a way to take all of your information and put it in one place because what happens now is if you if you get say a a new prefix from you buy it on the the open market you have to go to each of your upstream providers you have to provide a letter of authorization that says I'm allowed to advertise this block please please add this into your filters what a routing registry does is it allows you to automate a lot of that and then with the routing registry you can also influence your traffic that way because you can say hey I'm receiving these these routing registry information without getting to to deep dive into it I'm receiving these so whatever I receive I'll just apply my my filters to so it automates a lot of it so when we're talking in IX you know an IX is a shared fabric so when you maybe get to a large data center something like that a lot of the large data centers have internet exchanges and basically all it is is everybody on the internet exchange is plugging into a shared fabric I used to use the term switch but it could be multiple switches so everybody is is plugged into the shared fabric and they exchange traffic directly across the fabric lower latency you you have a better handle on your traffic you you don't have to worry about hey two hops up is that portfolio overloaded your your your peering directly with a Netflix a Google the guy across the street kind of thing so this is where bgp meds can come into play like I was talking earlier this is one of the more likely scenarios where you're going to have access to you know the same pass multiple ways and what what many people see on an IX is if I can dump all of my Netflix traffic to an IX if I can dump all of my anything that's on the IX it's usually cheaper to do I can guarantee that hey I'm in control of my port so if I have one gig or 10 gig to the IX that's gonna be my bottleneck not you know two routers upstream so with our med we can say hey if I see you know if I seem Netflix here on the exchange or if I see provider B up here you know I don't want it to go to provider B because they may not be sending me routes on the IX or it may be cost me more to have traffic go to provider B so I'm gonna influence where my traffic goes to the i-x so when you're when you're talking IX stuff there's a couple specific things that are - - mikrotik one is the the next hop choice so on most IX is you're going to do for self and we'll will explain what that is in a second and then your max prefix limit so so why the force next hop self so most I X's have what they call transparent route servers these are they set in between you and the peers so here's us down here here's our here's our exchange fabric and here's our here's our different members on the on the exchange so with many of the I exes they have these transparent route servers so you establish a connection to the route server and you start announcing your prefix is to the route server everybody else who wants to do that they can join the route servers so if you have say a hundred people on the i-x you now have if a hundred people join the route servers instead of setting up a hundred different sessions you know to each individual person there now just setting up one or two most most I exes run redundant route servers so you're just setting up two sessions and you're getting everybody on the i-x and these transparent route servers they sit in the middle so they're they're like a bridge but they just reflect the routes back to you so that's where the next hop self because you're telling whoever your your peering with hey the next hop for you to get to me it's not this transparent route server it's me you know about me so when you send me traffic yes it goes the route server finds it but it doesn't go to the route server it goes from say Netflix to me so it forces the the other side to use a specific IP address to send the traffic to otherwise they want to send it to the route server because they're paired with the route server so when you're when you're pairing with an IX which is you know I'm a little biased but I think as a network grows an IX should be part of your strategy we're seeing anywhere from 40 to 80 percent of eyeball or ISP traffic that's access customers offloaded to five web sites and you guys can probably guess you know Netflix Google Facebook and a couple of the content networks so if you can you know put an IX in your strategy you're going to be have a better experience for your customers so then we talk about max prefix limit so when you're on an IX you don't know how many prefixes you're going to get really because you're you're not talking to everybody on the i-x you're talking to the route servers they're talking to the route servers so most I exes can can give you a ballpark hey set your max prefix limit to you know like this one set it to 150,000 that gives room for growth that gives a little bit of overhead but max prefix limit is a protection for you so what happens is if someone miss configures their BGP if they don't do remember me talking earlier if they don't do an explicit deny on their outbound I've I've done this and then I start shoving my whole routing table to one of my up streams and this is a method to kill that connection because you don't want bogus routes coming and poisoning your route table you don't want you know your own routes coming back to you you don't want routes that people are not supposed to be advertising to you most of the the denial of service that you hear about in terms of BGP is someone to hijacking a prefix you know someone hijacked all of my kurz Microsoft's prefixes and so now Microsoft is essentially wiped off the internet for anybody who's not doing proper filtering so this is a way to protect yourself on an IX when you don't know how many prefixes you'll be getting now if you're doing a appear with you know say say somebody that you're selling bandwidth to it's pretty easy hey I'm accepting to prefixes from you and denying everything else so I'm going to add in a protection that says my max prefix limit is 2 or 3 so if for some reason something happens on my side and I start getting 10 or a thousand prefixes I'm going to shut that connection down so we won't go through these these links too much these are here just for for you guys to kind of up your BGP game my good buddy Faisal from from Miami he's he's the one who put together this list these are mainly ways you can do a couple things with your your BGP whether you're single home multihomed whatever so I encourage you to to check these out Josh Haven Potter has a script that allows you to update an address list that says hey these are the known bad actors on on the Internet as far as them doing garbage when it comes to BGP there's there's quite a bit of others out there but you know these these will get you started and the thing I like about like fast NAT Mon here is in the mikrotik wiki there's instructions on how to do it so you can you could go home tonight and implement it so last thing is I'll have like my sanity check chain what I use for myself is up to about 57 57 different prefix or 57 different lines I'm going to put that on my blog for the my patreon subscribers so you can just cut and paste that you can edit it for your own information this presentation will be on the the mikrotik site a couple weeks or whatever you know I put a lot of links in here so you guys you're driving home flying home or whatever you can be thinking about this and then when you get home oh okay I want to go back and reference this so any any questions comments experiences yes sir if you only if you only have a single connection coming in as BGP useful for me it is but not not as useful with BGP you can do things like the communities so you can say hey if I'm getting a denial of service and my upstream provider supports it I can just start advertising whatever IP is getting attacked over my BGP community and it's it's automatic you know I don't have to call them up and say hey start blocking this to me I'm just advertising I'm telling them what the block you know via BGP I would recommend not not taking full routes you know just take a default route then when you add on your second provider you know maybe maybe do full change it to full routes or something like that but take a default route if your upstream providers still allows prepending is that preferable and if they don't allow it what is it depends on what their upstream provider does so you want to have a conversation with them and kind of do some research of hey who are your upstream providers because the the closer it is to your to your upstream provider where these BGP pre pins are stripped off or just plain dropped it doesn't do you as good but if they if they have an upstream that still you know recognizes pre pins or whatever then it gets a little handier the more you push that out now if you have multiple connections to the provider pre pins and they they accept pre pins that can that can work too you know instead of one of the tricks with BGP is if you have a say at 23 and you have two two connections from the same provider you can advertise 24 s out each provider and a 23 out each provider and the more specific path wins if you don't want to do that then you can prepend it and influence the traffic that way that way you can influence both inbound and outbound being a medium-sized ISP you at what point do you switch from a default route to a full table I like to switch as soon as I can because then I have a better view of the internet so when the more customers I have the more important that's going to be because now I can see hey these customers they're preferring provider B to get to Google they're preferring it better if I'm not taking those full routes it's just you know kind of round-robin or whatever other methods you're doing so if you're taking full routes you have a better view and and you have a better view to send that traffic and if they're if they're adding it into their BGP the rest of the Internet has a better view of you so I would say as soon as soon as you as soon as you get that second provider start start taking full routes or at least you know if you're like hey I I don't want to jump into that right away because maybe I'm going to do a router upgrade or something like that at least take a default route plus their customer routes like hurricane electric if you pier with them on an IX they will give you like a hundred and thirty thousand routes for free and that's all their customer routes so now you know okay is is getting to you know this destination over hurricane electric is that quicker than you know looking at a full routing table from kocha but when it comes to I mean your talk about manual intervention you know you see it better but as far as automatic is there any advantage immediately to the customer as far as I guess lower latency maybe lower latency better better pass so I I rely on hurricane and cogent and some of them to kind of know which pass to which providers are less congested so if I'm taking the full routes from them and you know hurricane is sending a community or something that says hey I'm going to artificially it's not really prepending but I'm I'm going to say this is a less preferred route if I'm getting full routes then I'm I'm taking in more information from my up streams and they're there doing the work for me so so I can say my router knows it just automatically knows what the best path is and so it it will just automatically take the best path so if hurricane has influenced that that path through their own traffic engineering I'm getting the benefit of a multi-million dollar company that has more traffic engineering than I do - to help me get get somewhere so you would see a benefit right away to follow up on that do you use the community strings that you get from the ISPs to influence your outbound routes is there any benefit to that yeah if if I if I know what those strings are and they they look good to me yeah I will I will route write filter rules that say hey if I see this community string or whatever you know do this with it I'm trying to think I think cogent cogent does this they send community strings that say hey these are North American routes these are European routes and then like I have a client that's on d6 in New York and so it is much better for them to say hey if it's a if I get a European route from Cochin I am going to set the local pref on that where I'm not going to prefer that I'm going to prefer it to d6 because I got a better connection to Europe over d6 so yeah if it's if it's something that it's beneficial yeah go for it all right thank you all for for listening thank you [Applause]

Channel: MikroTik

Views: 5,034

Keywords: mikrotik, routerboard, routeros, latvia

Id: zyy1I5k2-aU

Length: 50min 6sec (3006 seconds)

Published: Tue Apr 09 2019