WEBVTT

0:00:03.160000 --> 0:00:08.340000
 Hello and welcome this video titled,
 Influencing OSPF Path Selection.

0:00:08.340000 --> 0:00:16.220000
 So we're going to look at when OSPF has
 multiple paths to the same destination,

0:00:16.220000 --> 0:00:21.360000
 we're going to answer questions like,
 is there a way to control how many

0:00:21.360000 --> 0:00:24.360000
 multiple paths it can install
 in the routing table?

0:00:24.360000 --> 0:00:28.540000
 Is there a way to control which path
 it selects as the best path?

0:00:28.540000 --> 0:00:32.480000
 If I don't like the one it's selecting
 as the best path by itself, can

0:00:32.480000 --> 0:00:36.560000
 I change that? We're going to
 look at that in this section.

0:00:36.560000 --> 0:00:41.800000
 Okay, most routing protocols,
 BGP is the exception.

0:00:41.800000 --> 0:00:47.080000
 Most routing protocols, if they learn multiple
 paths to the same destination,

0:00:47.080000 --> 0:00:51.380000
 and if those paths are equal, the metric
 is the same, we call that equal

0:00:51.380000 --> 0:00:55.040000
 cost routing. And most routing protocols
 will say, okay, if those two

0:00:55.040000 --> 0:00:59.440000
 or three paths are all the same to network
 50, 50, I'll install them all

0:00:59.440000 --> 0:01:03.000000
 in my routing table, then just
 load bounds between them.

0:01:03.000000 --> 0:01:10.700000
 So most routing protocols allow up
 to four equal cost paths into your

0:01:10.700000 --> 0:01:15.680000
 routing table. You can use the maximum
 dash paths command to increase

0:01:15.680000 --> 0:01:19.420000
 or decrease that amount
 if you so desire.

0:01:19.420000 --> 0:01:23.980000
 All right, another thing.

0:01:23.980000 --> 0:01:33.480000
 We know that OSPF uses as its formula
 as their review, 100 million divided

0:01:33.480000 --> 0:01:43.080000
 by the interface bandwidth in bits
 per second, which means if we have

0:01:43.080000 --> 0:01:48.540000
 a fast ethernet interface, a gig interface,
 a 10 gig interface, or anything

0:01:48.540000 --> 0:01:52.900000
 even higher, they will
 all equal to one.

0:01:52.900000 --> 0:01:58.400000
 They'll all have the exact same cost,
 which leads us to this next bullet

0:01:58.400000 --> 0:02:01.440000
 point. For networks with gigabit ethernet
 and above, this can lead to

0:02:01.440000 --> 0:02:04.560000
 inferior path selection.

0:02:04.560000 --> 0:02:10.920000
 So we can use this command right
 here to change that formula.

0:02:10.920000 --> 0:02:15.940000
 Maybe I instead want 1 billion divided
 by the bandwidth in bits per second

0:02:15.940000 --> 0:02:19.460000
 or 100 billion divided by the
 bandwidth in bits per second.

0:02:19.460000 --> 0:02:23.760000
 So by using the auto dash cost reference
 dash bandwidth, we can change

0:02:23.760000 --> 0:02:26.700000
 that initial number on
 the top of the formula.

0:02:26.700000 --> 0:02:29.740000
 Like this says here on the bottom,
 if you're going to do this, you're

0:02:29.740000 --> 0:02:32.920000
 changing the way OSPF comes
 up with its metric.

0:02:32.920000 --> 0:02:36.380000
 So all the routers should be using
 the exact same formula or they will

0:02:36.380000 --> 0:02:42.100000
 have an inconsistent view of what the costs
 are to reach particular destinations.

0:02:42.100000 --> 0:02:51.600000
 Okay, I showed you in a previous video
 that we can actually modify cost

0:02:51.600000 --> 0:02:53.360000
 in one of two ways.

0:02:53.360000 --> 0:02:58.000000
 We can modify it by changing the bandwidth
 on an interface, artificially

0:02:58.000000 --> 0:03:02.640000
 setting a bandwidth number we want,
 or by changing the cost value itself

0:03:02.640000 --> 0:03:04.240000
 on the interface.

0:03:04.240000 --> 0:03:08.880000
 Either way, will influence OSPF cost,
 either to go up or to go down and

0:03:08.880000 --> 0:03:13.080000
 make that path either less or more
 preferable than it previously was.

0:03:13.080000 --> 0:03:19.860000
 And the last thing I want to talk about
 is how to originate default routes

0:03:19.860000 --> 0:03:28.940000
 in OSPF. So OSPF is quite a bit different
 in how you get to input to inject

0:03:28.940000 --> 0:03:33.700000
 a default route than other routing
 protocols like RIP or EIGRP.

0:03:33.700000 --> 0:03:37.600000
 So with OSPF, it actually has a special
 command as you can see here just

0:03:37.600000 --> 0:03:42.400000
 for this purpose, the default dash
 information originate command.

0:03:42.400000 --> 0:03:44.340000
 Let's just talk about this
 here for just a moment.

0:03:44.340000 --> 0:03:46.000000
 Let's look at this example.

0:03:46.000000 --> 0:03:50.480000
 So in this example, we have this router
 right here, and maybe he's doing

0:03:50.480000 --> 0:03:55.780000
 BGP with an internet router.

0:03:55.780000 --> 0:03:59.340000
 Now, with BGP, you can actually literally
 exchange hundreds of thousands

0:03:59.340000 --> 0:04:01.340000
 of routes with a peering router.

0:04:01.340000 --> 0:04:05.240000
 And as of this time of this recording,
 there are literally, I think, over

0:04:05.240000 --> 0:04:09.400000
 800,000 routes consisting
 of the internet right now.

0:04:09.400000 --> 0:04:13.580000
 That's how big your BGP table could
 be, especially if your router that's

0:04:13.580000 --> 0:04:18.120000
 owned by AT&T, sprint, time-owner
 cables, something like that.

0:04:18.120000 --> 0:04:23.160000
 Now, let's say that router X here is
 learning of a bunch of BGP routes.

0:04:23.160000 --> 0:04:28.380000
 Now, router X, he is the only way to
 get out to the outside as far as

0:04:28.380000 --> 0:04:30.260000
 the OSPF domain is concerned.

0:04:30.260000 --> 0:04:32.400000
 Here's our OSPF domain.

0:04:32.400000 --> 0:04:38.240000
 Or maybe this. Maybe router X
 isn't learning any BGP routes.

0:04:38.240000 --> 0:04:38.940000
 He does like this.

0:04:38.940000 --> 0:04:40.460000
 He does has a static route.

0:04:40.460000 --> 0:04:43.640000
 He's got a static route saying, hey,
 to get to the default, to get to

0:04:43.640000 --> 0:04:47.600000
 anything I don't know, send
 it to my internet gateway.

0:04:47.600000 --> 0:04:49.200000
 Send to my ISP routers.

0:04:49.200000 --> 0:04:51.100000
 That's probably more feasible.

0:04:51.100000 --> 0:04:55.460000
 So now what we want to do is we want
 to have router X generate a link

0:04:55.460000 --> 0:05:03.420000
 state update. And inside of it, we want
 him to generate an LSA for a default

0:05:03.420000 --> 0:05:10.040000
 route. This is where we would use the
 default information originate command.

0:05:10.040000 --> 0:05:14.160000
 Now notice, there's also a keyword at
 the end of here called always, which

0:05:14.160000 --> 0:05:16.960000
 is optional. You don't
 have to type in always.

0:05:16.960000 --> 0:05:19.320000
 When would you use that
 and when would you not?

0:05:19.320000 --> 0:05:24.140000
 Well, if the router, like router X in this
 case, that's going to be advertising

0:05:24.140000 --> 0:05:26.160000
 the default route into OSPF.

0:05:26.160000 --> 0:05:31.160000
 If he already has a default route, learn
 via some other method, a default

0:05:31.160000 --> 0:05:34.800000
 route via static route, like we see here,
 a default route that you learned

0:05:34.800000 --> 0:05:37.780000
 via EIGRP or BGP.

0:05:37.780000 --> 0:05:41.900000
 If he's already got a default route, then
 all we have to type in is default

0:05:41.900000 --> 0:05:44.200000
 information originate.

0:05:44.200000 --> 0:05:47.600000
 And then he'll go ahead and send out an
 LSA to the rest of the OSPF routers

0:05:47.600000 --> 0:05:49.140000
 of a default route.

0:05:49.140000 --> 0:05:52.780000
 Now, what if he doesn't have
 a default route himself?

0:05:52.780000 --> 0:05:57.200000
 Like in my first example, when we didn't
 have the static route back when

0:05:57.200000 --> 0:06:01.080000
 you said, hey, these two guys here
 are here, they're doing BGP.

0:06:01.080000 --> 0:06:08.400000
 And this guy is sending 800
,000 plus routes to router X.

0:06:08.400000 --> 0:06:13.340000
 So router X, he doesn't need a default
 route because he knows everything.

0:06:13.340000 --> 0:06:15.540000
 He knows all the routes that
 exist in the entire world.

0:06:15.540000 --> 0:06:17.780000
 So he doesn't need
 a default route.

0:06:17.780000 --> 0:06:21.620000
 So if on router X, if I did default
 information originate and then I hit

0:06:21.620000 --> 0:06:25.280000
 my return key, you would
 see it doesn't work.

0:06:25.280000 --> 0:06:29.240000
 Because that command only works if
 the router has a default route that

0:06:29.240000 --> 0:06:34.700000
 you've learned via some other method,
 unless I add the keyword always.

0:06:34.700000 --> 0:06:38.400000
 So you got to be real careful with this
 keyword because when you say default

0:06:38.400000 --> 0:06:43.400000
 information originate always, that means
 you will send out a default route.

0:06:43.400000 --> 0:06:47.700000
 You'll tell everybody, hey, come to
 me with your unknown traffic, even

0:06:47.700000 --> 0:06:51.120000
 if you don't have a default
 route yourself.

0:06:51.120000 --> 0:06:54.520000
 Now, like I said, if router X has knows
 about everything, he doesn't need

0:06:54.520000 --> 0:06:56.220000
 a default route.

0:06:56.220000 --> 0:07:01.140000
 But router X could potentially say default
 information originate always

0:07:01.140000 --> 0:07:05.800000
 and send out a default route when
 he himself is missing information.

0:07:05.800000 --> 0:07:08.400000
 And that would be problematic.

0:07:08.400000 --> 0:07:16.780000
 So that concludes our refresher on
 OSPF theory and implementation.

0:07:16.780000 --> 0:07:17.420000
 Thank you for watching.

