网络课程笔记

00:00 Hello, and welcome to this lecture,

00:01 Dynamic Routing using Border Gateway Protocol.

00:04 So in the earlier lecture,

00:05 we understood the difference between the static routing

00:08 and the dynamic routing.

00:09 And we understood that in case of static routing,

00:11 you have to modify the route entries in your route table,

00:14 and you have to add the new routes

00:15 that you want to send traffic to manually.

00:19 And we also understood that in dynamic routing,

00:21 these routes are propagated automatically.

00:24 Now in this lecture,

00:25 we are really going to understand how that propagation

00:27 happens and for doing that,

00:29 you need to use a different protocol in this case,

00:32 and that would be a border gateway protocol.

00:34 Now border gateway protocol is a backbone of internet as

00:38 well because you know,

00:39 internet is all mesh off a lot of networks.

00:41 And then all these networks needs to exchange the dynamic

00:45 routes with each other,

00:46 and that's where you can hit any IP address

00:49 and your packet reaches to the destination.

00:51 Now, border gateway protocol is also important

00:54 with respect to your exam.

00:56 And in case of AWS,

00:58 both AWS side to side to VPN, as well as direct connect,

01:02 both support the border gateway protocol.

01:04 So let's understand that.

01:05 And in that case let's get started.

01:08 Okay, so what is border gateway protocol or in short PGP?

01:13 It's a dynamic routing protocol using path-vector.

01:16 Now what that means is,

01:18 in case of border gateway protocol,

01:20 there is something called AS, autonomous systems,

01:23 and we learned about that in the last lecture.

01:25 So any network,

01:26 you can call it as like AS

01:28 and this AS is interchange

01:30 the routing information between each other.

01:32 So in case of BGP, they work on path-vector protocol,

01:36 which means they exchange the best part

01:37 to reach to the destination.

01:40 So if I know the parts of other AS's,

01:43 how to reach them, what is the next stop?

01:45 I will exchange that information with my neighbor BGP,

01:49 and that's where the routes gets exchanged.

01:52 Now, there are two types of the BGP.

01:54 One is iBGP, or you can call it internal BGP.

01:58 So it is like routing within one autonomous system.

02:02 And what we are interested in, in eBGP,

02:05 which is routing between the different AS's

02:08 because we are connecting multiple networks, right?

02:11 And that's where the routing would happen between

02:13 two different AS's.

02:15 So that's called eBGP.

02:17 Now, how is this routing decision is influenced?

02:19 Now what that means is,

02:21 if I want to reach to a destination,

02:24 what parameters I need to consider.

02:26 Can I influence the routing?

02:29 Suppose there are multiple

02:30 routes to reach to the same destination.

02:32 So through BGP can influence that I want to take one route

02:35 or the other.

02:37 Now there are certain parameters or features,

02:40 or you can say attributes using which you can influence this

02:43 routing mechanism.

02:45 So one is called weight.

02:46 So you can define the weight on the different paths,

02:50 and then the routing would be taken to the highest weight,

02:54 but that only works in the Cisco routers.

02:56 And that's where we are not really going too deep into

02:59 this weight attribute.

03:01 But there is something called AS PATH,

03:03 which is like how many hops are there

03:05 to reach to the destination.

03:07 Ultimately that is called ASPATH.

03:09 So ASPATH would contain the number of the hops or which

03:12 AS's I need to cross in order to reach to the destination.

03:16 So we'll see how AS PATH influences that.

03:18 Then there is something called local preference.

03:20 Now this works within the AS.

03:23 And we'll talk more about that.

03:25 So if you have multiple exit points to reach to the target

03:30 destination from your source network,

03:32 then within your source network,

03:34 which router you will choose to go to that destination,

03:36 that is basically a local preference.

03:40 And then there is something called MED,

03:41 Multi Exit Discriminator.

03:43 Now similar situation, like a local path.

03:46 But in this case, the MED value is external.

03:51 That means the values are exchanged

03:53 between the AS's as well.

03:55 And it will be more clear as we see the scenario,

03:59 how it works.

04:00 So let's try to understand how everything works in the BGP.

04:05 So let's assume you have the whole world at your disposal.

04:09 And we are now trying to set up a network connectivity

04:12 between the data centers at various locations.

04:16 So there is one network you can call it one AS and it has

04:21 IP range ten, ten, zero, zero, zero, 16

04:23 and AS value as 100.

04:26 Now you understood that there is private AS and public AS,

04:29 anything works here.

04:30 But we are taking the number as 100.

04:32 However, the actual number, you know,

04:34 there is some range that you can look at

04:37 in the last lecture.

04:39 Similarly, there are other networks like one maybe Africa,

04:43 and which has AS 200

04:46 with a CID arranged 10, 20, zero, zero, 16.

04:49 And then there is somewhere in India

04:52 and maybe somewhere in the Europe.

04:54 So we have four networks here

04:55 and we have given them different AS numbers: 100, 200,

04:58 300 and 400.

05:00 Now every AS would be consist of multiple routers.

05:03 That's what is shown here.

05:05 And then there will be one router,

05:07 which is connected to other networks,

05:09 but that router has a private IP address from the network

05:14 address range, and that's what is shown here.

05:17 Now what we have done is,

05:18 we have connected this network or

05:21 either a MPLS link or other links.

05:25 And there is a connectivity between AS 100 and AS 400,

05:30 similarly between 400 and 300, 300 and 200,

05:33 and then 100 and 200.

05:35 So all these are connected

05:37 and there is a different network

05:38 bandwidth that they are connected with,

05:41 and we will show you why that is so, as an example.

05:45 And additionally,

05:46 there is one more link between AS 400 and 200.

05:49 So this is how the network is laid down.

05:52 And now let's look at how BGP works in this scenario.

05:56 In terms of the BGP neighbors.

05:58 As I said,

05:59 if the traffic is going between AS's, different As's,

06:04 they are called E BGP neighbors.

06:06 So these two guys are E BGP neighbors versus,

06:09 if you're talking about router within the same AS,

06:12 and they exchange the routing information,

06:14 they are called I BGP neighbors.

06:16 And as I said,

06:16 what we are interested in is E BGP neighbors.

06:20 Okay, so this is a network and now let's see

06:23 how the routing information is exchanged.

06:25 So every AS would have their own routing table.

06:30 We also call it BGP table.

06:33 Now in that BGP table,

06:34 there will be three things: the destination, the next hop,

06:37 and the path, you can call it AS PATH.

06:40 Now if you look at this AS,

06:41 the destination is itself ten, dot, ten, dot, zero, zero 16.

06:45 The next hop in that case would be zero, zero, zero,

06:48 and the path would be I, I for internal.

06:50 That means all this traffic is internal traffic.

06:53 Now, every AS at initial level,

06:56 it would have this kind of route table where the entry

06:59 would be for itself.

07:01 The CID actually will be the self CID address.

07:04 Now what happens, as you know this,

07:06 all AS's are connected to each other.

07:08 They learn the route from the other AS,

07:12 like if from AS 100, I want to reach to AS 400,

07:16 then what route entry I should have.

07:18 So they would exchange those routes

07:20 and that's where all these route entries are here.

07:24 Now let's take an example from AS 100,

07:26 if I were to go to AS 200

07:28 which is this one, with the destination IP address of

07:31 10, 20, zero, zero, 16, the next hop is this router,

07:35 which is 10, 20, zero, one.

07:36 And in this case, my AS path would be 200.

07:39 So I'm adding that 200 and there will always be this

07:42 internal entry.

07:44 So AS PATH is 200 comma I.

07:46 Now if I'm to reach to 400, of course,

07:48 my next destination is 10, 40, zero, one,

07:52 which is this router.

07:53 And the AS PATH would be 400 comma I.

07:56 Likewise, we added

07:58 or it gets automatically added all the AS entries

08:01 out table entries for all other AS's.

08:04 Now if you see specially, this AS 400 and 200,

08:07 they're connected to rest of the three AS's.

08:09 That's where there are self entry and three more entries for

08:12 all these three AS's.

08:14 So this is very basic to start with.

08:16 The route table would have this kind of entries.

08:19 And now if you can see, if I want to reach to AS 400,

08:22 any IP address 400, it will take that path.

08:25 And if I were to go to AS 200 from AS 100,

08:29 it will take this path.

08:31 Now, what if I want to reach to ASA 300 IP address?

08:36 Now, if you look at this route table,

08:37 there is no route entry for that.

08:39 That means right now,

08:40 I can't reach to that destination.

08:42 And that's where the BGP comes to the rescue, right?

08:45 What it does is,

08:47 whatever path you or other neighboring AS's understand

08:51 or learn, they will pass that counting information

08:55 to other AS's.

08:56 So in this example,

08:57 if you see my AS 400 knows how to reach to AS 300

09:03 using the hop 10, 30, zero, one, right?

09:06 What it does, it just upends its own AS number here 400,

09:12 and sends this information back AS 100.

09:15 Now, if you see AS hundred got the destination address,

09:19 10, 30, zero, zero, 16.

09:21 And in this case, the next hop is 10, 40, zero, one

09:25 because this directly does not understand the,

09:28 it does not have direct connectivity with AS 300.

09:32 So next hop will be a 10, 40, zero, one

09:34 and as packet reaches this destination,

09:37 it knows how to route traffic to AS 300.

09:41 So it will learn that entry here.

09:43 Similarly, it will also get the same entry or similar entry

09:47 from AS 200 as well.

09:49 That's because AS 200 also connected to AS 300

09:53 and it has its own entry.

09:55 So now, AS 100 has multiple destination entries for AS 300.

10:02 And that's what the BGP is.

10:04 What it does, it exchanges routing information with all

10:07 neighboring AS's.

10:09 And in this case, we are just showing AS 100 path,

10:12 but this AS 400 will also receive similar entries

10:15 from AS 100 and likewise.

10:18 So it will be a lot of route getting exchanged

10:20 between all these AS's.

10:22 Now what's the benefit of this?

10:23 Like AS 100 has now two entries to reach to AS 300.

10:27 So what would happen?

10:29 Suppose by default BGP chooses the,

10:33 this path that is through AS 400 to reach to the AS 300

10:37 and somehow this link goes down,

10:39 that means there is some problem with that network.

10:43 Now, BGP can't take other path because in routing table,

10:46 it knows that there is one more path exists to reach to

10:48 the target destination.

10:50 And that's where BGP can be used for fail-over

10:54 kind of scenarios.

10:55 So that's one of the possibilities and we are going to talk

10:57 more about such scenarios.

10:59 But what I want you to understand from this

11:02 is that a BGP is all about exchanging

11:05 this routing information dynamically.

11:08 Now what do I mean by dynamically?

11:09 Suppose tomorrow this AS 400 gets connected to a new,

11:13 say, AS 500 and then it would learn about that route

11:18 and there will be a route table entry

11:20 in the AS 400 route table.

11:22 Now it would also publish that route table information to

11:25 AS 100 and AS 200 and AS 300.

11:28 So that now all these neighbors also understand about

11:31 the new network,

11:33 which got connected to AS 400 and that's where they can

11:36 route the traffic to, even that destination.

11:38 So BGP is all about that.

11:40 Internet is all about that.

11:42 Okay, so that's basics of BGP.

11:45 And now let's get into the attributes of the BGP

11:49 that we talked about.

11:50 Specifically, the AS PATH, local preference,

11:54 and then multi exit discriminator.

11:57 So stay tuned.