Implementing Cisco IP Routing (642-902 ROUTE) Tutorials
BGP / Overview of BGP pt. 2
Visitors to VTC.com will be able to view all introductory videos for each training course.
Free Trial Members will gain access to first three chapters for each training course.
Full Access Members have full access to VTC.com�s entire library of video tutorials.
Learn More
Subtitles of the Movie (Implementing Cisco IP Routing (642-902 ROUTE) / BGP / Overview of BGP pt. 2)
So we're discussing an overview of BGP and we've just given the briefest of overviews of how AS routing works and again this will make more sense once we put this in a lab, as most things do, when I put them in a lab, because if you're like me, again, you learn better when you actually see it in production. So obviously BGP is different than IGPs or Internal Gateway Protocols and we'll take a couple of minutes to look at these key differences. First off, BGP requires manually configured neighbors. It's not like OSPF or EIGRP or really any other routing protocol where if you set it up on a Point-to-Point link, it'll just say, hey there's OSPF neighbor over there, he talks OSPF, I'm going to start exchanging routes with him. No I really don't quite work that way and, and with good reason. Let's say you're an ISP and you're running BGP internally to, you know, route to your Internet providers, well you have a customer who sets up their own BGP router and starts injecting routes into your BGP domain, you can, you can really mess up the Internet if you, if you do that wrong. In fact, if you'll recall a few years ago and I record this at the end of 2011 but a few years ago, there was a, a period of time where YouTube just disappeared from the Internet for, you know, a few hours, two or three hours I think. You tried to go to YouTube and it just didn't work and that was because there was some ISP in the Middle East, I think it was Iran or Syria that started advertising the network blocks that belonged to YouTube as belonging to them and they advertised them with a more specific network mask and you'll recall basic routing, the longer the network mask, ie the more specific route always wins, so, you know and I'm just pulling numbers out of the air, but let's say YouTube has several Class B network addresses. Well this rogue ISP started advertising those routes into BGP as Class C prefixes. Well the entire Internet said, well, you know, heck that's the more specific route, I'm going to prefer that. So it started routing all the traffic over to this rogue ISP. Obviously that's bad for the Internet health in general and that's why with good reason you have to manually configure your neighbors on both ends before the BGP neighbors will establish. And at this level perhaps getting the neighbors to establish in BGP is one of the more tricky tasks we'll do, just because there are so many rules that you have to take into account. You know, this neighbor has to, you have to have the Multihop set to a certain number so that you can get to the loopback and all of these rules and again we'll go over that in the lab, that, all that, long story short, you have to have manual neighbor setup in BGP. Also BGP uses TCP, uses TCP Port 179 to be specific. It is not it's own Layer four protocol, like OSPF and EIGRP, they use IP obviously to send the Routing Updates, but they are their own Layer four protocol. BGP uses TCP and as a result, you don't have to worry about making sure that your Route Updates made it to the other router. You don't have to worry about building an acknowledgment into BGP at least a the Layer three level, you let TCP handle that. So basically you're just sending data like you would with just about any other Layer 4 protocol. BGP uses what's called Path Attributes to select the best route. Where most other IGPs use the Metric, obviously OSPF is just Bandwidth divided by 100 and you've got that big long formula for EIGRP and RIP uses hop count, BGP uses Path Attributes to select the best route. Now I'll take a little bit of time and talk about Path Attributes. Let's say that we have a route to say the 10.0 Networks. We'll write 10.0 and you have a BGP router and it's got two routes to get to the BGP network, like so. Well BGP Tags both of these routes with all of these Route Attributes and there's lots and lots and lots of them. We'll look at how BGP determines the best path using these Path Attributes in another video but basically the BGP router says, okay, well I've got these two paths to get to the 10.0 network. I'm going to start comparing these Path Attributes. Is this the same? Yes that's the same. This, is this the same? Yes that's the same. Is this Path Attribute the same? Oh wait, this Path Attribute is different, I'm going to prefer this route to get to the 10.0 network, I'm just going to throw this one in the garbage. So rather than just computing a big Metric and it does use a Metric in the background once it goes through all these Path Attributes, rather than just using a single Metric, BGP uses these Path Attributes to determine what is the best path to the destination network. So I'll clear off all of my scribble there. The next difference is that it prefers Scalability over Fast Convergence. Obviously as we've discussed, the Internet is huge and imagine what would happen, you know, like say RIP sends it's updates every 30 seconds. Well you've got this huge Route Table and imagine if the BGP Route Table was sent to every router in the BGP domain every 30 seconds. Well that's you know, you'd, you'd have constant traffic because you've got a Route Table that maybe 30, 40, 50, 70, 100 megs, imagine sending that out every time a link went up or a link went down. It would just be unwieldy so BGP rather than being a very fast converging protocol, it prefers Scalability where you can have millions of routes in your Route Table or in your BGP Table and you don't necessarily converge on a dime. As a result, BGP is the slowest converging routing protocol on the planet. You'll see this in the lab where we set up our neighbors and then we just got to sit and we got to wait for 60, 90 seconds until the neighbors establish and then you'll put routes into the BGP, you'll start injecting routes into BGP and you'll have to wait 60 seconds for those routes to show up, even in the local router and then you got to wait another amount of time before they'll make it out to the destination router on the other end of the network. It can make troubleshooting really interesting, because you're sitting there and you're expecting it to be like OSPF or EIGRP where it's nearly instantaneous and it's not. One last thing to remember is that BGP will use the Autonomous System Path, Path Attribute as the default discriminator or the default tie breaker unless you tune it properly. Remember we said it acts just like RIP unless you tune it? Well this is how it acts like RIP. Basically it uses hop count and that's what the AS Path, Path Attribute is. We'll continue with our overview of BGP in the next video.
Tutorial Information
| Course: | Implementing Cisco IP Routing (642-902 ROUTE) |
| Author: | Greg Dickinson |
| SKU: | 34291 |
| ISBN: | 978-1-61866-028-2 |
| Release Date: | 2011-12-28 |
| Duration: | 10 hrs / 105 lessons |
| Work Files: |
Yes |
| Captions: | No |
| Compatibility: |
Vista/XP/2000, OS X, Linux QuickTime 7, Flash 8 |
VTC Sign up & Benefits
- Unlimited Access
- 81,350 Video Tutorials (14,200 free)
- Video Available as Flash or QuickTime
- Over 715 Courses
- $30 for One Month Access
- Multi-User Discounts Available
