5. 11.4 RIP
One of the dynamic routing protocols you need to know about for the Network Plus exam is Rip, which stands for Routing Information Protocol. And there are a few different versions I want you to know about. First up is Rip version one or RIPV one. By the way, I would not use this on a modern network. I actually did use this on a network at a university way back in the early 1990s. But I would would certainly not use it today. It’s got some undesirable characteristics. For example, the way it sends out routing information to other routers is through broadcasts. Now remember, a broadcast goes to every device within a subnet or within a broadcast domain.
That means when the router says, here’s my routing table, I can get to all of these networks. That advertisement goes not just to neighboring routers, it goes to PCs and printers and every other device in that subnet. And to compound that, it happens by default every 30 seconds. It’s like me coming up to you and saying, hey, let me read you my resume of everything I know how to do, and I read this resume and I come back 30 seconds later and I do it again. And 30 seconds later and I do it again. That’s a bit much. And another downside to Rip version one is it does not have what is called VLSM support for your notes. That stands for variable length subnet masking.
What that’s telling us is that rip version one can only advertise classful networks. Meaning that if we go in and add some bits to a subnet mask to take one network and divide it up into different subnets, if we do that, we cannot advertise those networks that we created by adding bits. We have to advertise at that class full boundary. So let’s say you take the ten one 10 network and you’ve applied a 24 bit subnet mask. If you tell Rip to advertise that, it’s going to say, no, I’m just going to advertise the classful address for a ten network which has eight bits in the subnet mask. Rip version one also only supports IP version four. Now, things did get a lot better with rip version two. Instead of sending its advertisements everywhere, a rip version two router is a bit more selective.
It’s going to use a multicast which is going to go just to rip version two speaking routers because those routers join a multicast group with a group address of 2240 nine. This version does support variable length subnet masking. It also adds authentication. So you cannot have someone just bring up their own router and start exchanging routes and maybe injecting some false routes that could happen with rip version one. But like Rip version one, we still advertise by default our entire network table every 30 seconds. And it only supports IP version four. There is a version that will support the writing of IP version six.
And it’s called Ripng, where Ng stands for Next Generation. And if you’re a Star Trek Next Generation fan, yes, they did get the name for Ripng from Star Trek The Next Generation. This will support the routing of IP version six, and the multicast address that it sends its advertisements out to is the IP version six multicast address of FF two nine. And the way Rip selects a best path to a network, regardless of which version we’re talking about, is through the use of hop count. How many routers do I have to go through besides myself to get to this destination network? If a network is directly connected to me, then it’s zero hops away. If I have to go through one router to get to that network, it’s one hop away. And Rip is not very scalable because it can only go through 15 router hops. It sees 16 router hops as infinity.
It’s unreachable. And we mentioned that it’s going to send out its entire table by default every 30 seconds. That’s called a full update. But we can have triggered updates if there is a change in the routing table. We can go ahead and advertise to our neighbors that change to the routing table without waiting for that 32nd interval. And I also want you to know about a couple of protection mechanisms that Rip can use split Horizon and Poison Reverse. In fact, let me give you an animation of what that’s doing for us. Here we see a fairly simple network. We’ve got three routers, and we’re just looking at the routing table of routers R two and R three. Let’s make sure this makes sense to us. First of all, router R two is directly connected to the 170 216, 100:24 network. So in R two’s routing table, that network shows as zero hops away because we’re directly connected. Same thing for the 192 dot 168 dot one dot zero size 24 network. It’s zero hops away. But take a look at ten dot one, dot one dot zero size 24 that was advertised to us by R three. That network is over on the right hand side of R three. That’s where switch SW one lives. R three is connected to it, so it’s zero hops away from R three. We can see in R three’s writing table, but when we advertise it over to R two, r three says, yep, I’m directly connected, I can get you there. And R two says, great, my hop count to ten 10 size 24 is one hop. I have to hop through R three to get there. R three’s writing table, as we’ve already said, has a zero hop count for ten one 100:24. It’s also directly connected to 170 216 100:24. So that’s a hop count of zero.
And R two is advertising to R three. That network between R one and R 2192, 168, 100:24. So that’s one hop away from the perspective of R three. Now, let’s say that something horrible happens to SW One, that network goes down. That interface is no longer up on R Three. That network is not available. If we did not have the protection mechanisms of things like Split Horizon and Poison Reverse, here’s what could happen. If Router R Two sent out every 30 seconds its entire routing table to all of its neighbors. Here’s what happens. R Two is going to advertise its routing table over to R Three, including the advertisement for Ten One is going to say to R Three, hey, I can get you there, and I’m only one hop away. And R Three says, oh, that’s great, because I just lost my connection. I’m glad you can get there in one hop. So I’ll go through you. And from my perspective, it’s two hops. Here’s the problem. The reason R Two thought that network was one hop away is because R Three told it it was directly connected to it. And R Three has lost that connectivity.
So these routers have a false sense of security right now. And after the advertisement interval for R Three rolls around, it’s going to advertise back to R Two. It’s two hops away from Ten one 10. R Two gets that, and it says, the guy that was telling me they were one hop away now seems to be two hops away. Okay, that makes me three hops away. And then we advertise that back to R Three. And then R Three says, I must be four hops away. And it goes on and on and on until we get up to 16 hops, which is considered to be infinity. To prevent that from happening, Rip can use a mechanism called Split Horizon. Split Horizon is going to say, if I, as a router, receive a routing advertisement in on a specific interface, I will not advertise that same route out of that interface.
In this case, router R Two received the advertisement for Ten One 100:24 in on its interface on the right. It’s not going to advertise that back out. With split horizon in effect. And there’s yet another protection mechanism that Rip can use. After that network where SW One is goes down, r Three is going to say, I’ve lost connectivity with that network. I’m 16 hops away. Let’s take that out of my routing table. But it’s going to make sure that R Two no longer relies on it to get to that network. It wants R Two to know that that network is no longer available. So it’s going to send a poisoned route advertisement to R Two saying, hey, that network ten 100:24, it’s 16 hops away. When R Two gets that, it says, uhoh, 16 in my book. That’s infinity. That’s unreachable.
So R Two is going to remove it from its routing table. That’s another way of eliminating that false sense of security where R Two still thinks there’s a way to get to Ten One 10. And again, that’s called poison reverse. That’s where a router that knows a network has gone down. It can send an advertisement with an infinite metric to its neighbor for a route that it knows is unreachable. So r three sent an advertisement to r two with an infinite metric for ten, one 10. And sometimes you might hear rip called routing by rumor, because that’s really what’s happening here. One router tells somebody else what they know, and they tell somebody else what they know. And that’s a look at rip, the routing information protocol.
6. 11.5 OSPF
For years, I’ve asked my students what routing protocol they use on their corporate networks, and overwhelmingly, the response has been OSPF openshortest path first. OSPF is one of our link state routing protocols. And the link state routing protocol has a map of the network. It knows how each of the routers within an area of the network are interconnected. Unlike EIGRP, that knows, here’s the next hop to get to the destination. Eichrp doesn’t know how the routers are interconnected, but OSPF does. And OSPF is one link state writing protocol you might run into out in the world, but another link state writing protocol that I want you to be aware of, not covered in this video. And that’s intermediate system to intermediate system. So we’ve got OSPF and ISIS. And as a metaphor, I want you to think of OSPF much like working a puzzle. If you and your friends are sitting around a table working a puzzle, you have puzzle pieces, they have puzzle pieces.
You each have portions of the overall picture that you’re trying to put together. That’s much the way that OSPF works, different routers are connected to different networks within this area, and they can share their information or metaphorically their opposite pieces with one another. And after they all exchange their information, they should have the same view of the network. And here are a few characteristics I’d like you to know about OSPF. It is an open standard, so it’s not vendor proprietary. And routers that are exchanging their puzzle pieces, their information with one another. They have formed adjacencies, and we can also form neighborships with OSPF.
And there’s a difference between an adjacency and a neighborship. I’ll distinguish between those two in just a moment. But we said these different routers know different pieces of information, and they’re going to collaborate together to put together this map of the network. Well, that information is in the form of LSAs link state advertisements. Those are sent between their routers to educate one another about what networks are available and how things are interconnected. And then within each router, those LSAs are grouped together. We put the puzzle pieces together to form the map of the network, and that map is referred to as the link state database.
And once we’ve got that database in place and we know how everything is interconnected, we know the bandwidth on each of the links between the routers, we can run the Dijkstra Shortest Path first algorithm, and it’s going to determine the optimum path between any one point of the network and any other point of the network. By the way, that is exactly how your car’s navigation system works. It also uses the Dexter algorithm to determine the shortest path. Or maybe you’ve got a navigation program on your smartphone, and your smartphone might do a recalculation. If there’s some sort of an accident on the road, it might detect that, and suddenly it assigns a higher cost to that roadway, and it can divert you around that.
Well, OSPF, if we have an issue in the network, we can reroute around that issue. If we do have a backup path, and once OSPF has built its link state database and it knows what it considers to be the best route to get to a specific network, that route becomes a candidate to be injected into the router’s IP routing table. But here’s the big point. Just because OSPF has a route to a network, it does not mean that that OSPF learned route will necessarily be placed into the IP routing table. Because we may have a route information source that still advertises that network. That’s more believable than OSPF. Maybe EIGRP is advertising the very same network. If that’s the case, then our router is going to believe EIGRP over OSPF, because EIGRP is more believable.
It has a lower administrative distance of 90 as compared to Ospf’s administrative distance of 110. And before we take a look at how things interconnect in an OSPF network, I want you to understand these terms. The first is a hello message. This is how those neighbor ships are formed, and we’re going to be talking about something in a few moments called a designated router. A hello message is also how we’re going to elect a designated router. And we said that the information that’s exchanged between our routers, that information is an LSA, a link state. Advertisement in our analogy, an LSA is much like a puzzle piece. However, here’s a big misconception. A lot of times people will say this router sent an LSA packet. Actually, that’s not technically correct. LSAs are information.
They’re not packet types. An LSU a link state update, that’s a type of packet that carries the LSA. So the packet is an LSU. The information inside of that LSU, that’s the LSA. And have you ever been working a puzzle and it seems like you’re missing a piece. You’re trying to work in this corner, and you ask your friends around the table, hey, does anybody have the piece that looks like this? I’m looking for a straight edge. You’re requesting a missing piece to the puzzle. You want to see if anybody else has it. That’s what a router can do. If a router has attempted to construct its link state database, but it’s missing a piece, it can request from other routers that missing piece of information.
Or in other words, that missing LSA, that’s an LSR, a link state request. And if a neighbor gives it that missing piece of information, it will say thank you in the form of a link state acknowledgment an LSAC. Now, I mentioned there was a difference between a neighborship and an adjacency. Let’s discuss that. First of all, a neighbor is with a router on our same network segment. We share the same subnet, and we’re going to exchange hello messages using multicast. We don’t broadcast it like Rip version one did we send a multicast hello to the multicast address of 2240 five. That’s with IP version four for your notes. For IP. Version six. It’s FF two. But this is what a neighbor is.
We’ve said hello to one another, but we’ve not exchanged information. That’s what an adjacency will do. And when I think about this, I think about my two neighbors. Where I live, there are only two neighbors anywhere near our house. Now, one neighbor I know pretty well, we’ve worked on some projects together. We’ve been to one another’s homes. We’ve got a much tighter relationship. We exchange information. The other neighbor, I really don’t know them very well, I know the car they drive. So I’ll wave at them as I’m leaving my driveway and we’ll say hello to one another. But that’s really it. We don’t have a very deep relationship. We just say hello. We’re just neighbors. In this scenario, my other neighbor that I do know better, that’s much like an adjacency.
Not only do we say hello, we say hello, but in addition to that, we exchange information. That’s what an adjacency does. Routers that are OSPF adjacent, they are neighbors. That’s a prerequisite. But in addition to just saying hello, they have exchanged information to build that link state database. And you might be wondering, it sounds like an adjacency is better. Why would we ever have just a neighbor when we could have everybody be adjacent with one another? Well, it’s not going to scale terribly well if we have, for example, an Ethernet segment, and this segment has multiple OSPF speaking routers on it, if everybody was adjacent with everybody else, that could be a lot of adjacencies. Consider this, we’ve got six routers. In this example, if every router were adjacent to every other router, that would be let’s do the math. Here’s the formula. It’s N times N minus one divided by two. So N, that’s the number of routers. That’s six. N minus one is five. So six times five is 30, divided by two is 15. We would have to have 15 adjacencies to fully mesh these routers together. And that’s with just six routers. Imagine if we had ten routers, that would be ten times nine divided by two. That’s 45 adjacencies. So what we can do instead on networks like this, they’re called OSPF broadcast networks because they all belong to the same broadcast domain. We can have designated routers. We can elect a router as a Dr or a BDR, a backup designated router in case the designated router is not available. And here’s the trick.
Once we elect designated and backup designated routers, the other routers in this network, they don’t need to form adjacencies with one another, they just need to form adjacencies with the Dr and the BDR. That dramatically reduces the number of adjacencies that we have to have. And I mentioned the multicast address that was used for routers to say hello to one another with OSPF here’s a reminder of those IPV four and IPV six addresses. But if we’re just wanting to communicate route updates with DRS and BDRs, it’s a different multicast address. For IP version four, it’s 2240 six. And for IP version six, it’s FF two colon colon six. That goes to all designated routers, which includes the backup designated router. Now, at this point, we’ve talked about how OSPF routers share a common view of the network.
They share a map. Well, technically, that map covers an OSPF area. Some networks only have one OSPF area, but you can divide up an OSPF area into multiple areas. That’s what we have here. We’ve got area zero at the bottom, and we’ve got area one and area two. And notice I’ve got a calculator next to each one. What I’m saying there is that the calculation, in other words, the Dijkstra algorithm is performed on each area. That way, if we have a really big area, that Dijkstra algorithm calculation doesn’t take more process or resources. And you might wonder how a network known to area zero gets advertised over to area one. Well, that’s the job of an ABR, an area border router. An ABR sits at the border of at least two areas, and it can send information back and forth between those areas.
Not an entire map. R three is not telling area one. Hey, here’s a map of area zero. It’s just saying, here is a list of networks available in area zero. You don’t have to run the extra algorithm on it, but if you want to get to any of those networks, come to me. I’ll get you there. That’s the job of an ABR. And notice that I’ve got an area zero. That’s actually a requirement. If you have more than one area in your OSBF network, you’ve got to have what is called a backbone area. And that backbone area is going to be numbered zero, or your areas can actually be numbered to look like IPV four addresses, that that one area could be named zero, dot zero, dot zero, dot zero.
Now, even though that looks like an IPV four address, it’s really not. We’re not saying here’s how to get to this network, or we might have area one’s area number as one one. We’re not saying we can get to that IP address. That’s simply the name of the area. You can write it either as just a decimal number, like zero, or a dotted decimal, like zero. I typically just use the decimal approach. And you might wonder, from a design perspective, when do we need to start adding areas? How big is too big? Well, a lot of people are too cautious about this, in my opinion. There was a recommendation that Cisco made many years ago, and that was if you have more than 50 routers in an area in order to not overburden those router processors when they’re running the Dixter algorithm, you might want to break that up into other areas. However, that recommendation was based on a very, very old Cisco router.
If you’re familiar with Cisco, it’s a 2500 series router, which is slow these days. The router speeds and processor capabilities are orders of magnitude faster than those old routers. So it’s not that big of an issue. Now, personally, I would not start breaking off my area into multiple areas just because I had maybe a couple of hundred routers in an area. Now, I may break off an area that happened to represent, let’s say, a data center. It might help me to do troubleshooting to have all of the data center routes in one link state database and my enterprise routes in a different link state database. Some might do that. And as one final consideration with OSPF, let’s think about how OSPF judges the best path. With Rip, we used the fewest number of router hops.
That’s not the case with OSPF. Consider this topology. We want PC One to communicate with PC Two, and it has two ways of getting there. It could go to R One, up to R Two, down to R Three and then out to PC Two. Or it could go to R One and go directly over to R Three. If we were running Rip, R One would say, oh, I can get to PC Two with one router hop. If I go through R three. I don’t want to go through two router hops. I’m certainly not going to go to R Two and then to R Three. I’ll just do it in one router hop. However, notice the link speeds, that link between R One and R three, it’s running at a measly ten megabits per second. The link speeds between R One and R Two and also R Two and R Three, those are 100 megabit per second links, so those are going to be faster.
Now, we can look at that and say, well, obviously I want to go over the 100 mega links. I don’t want to go over that ten mega link. That’s not the way Rip would look at it, though. Here’s how OSPF does that calculation. It makes its decision based on the cost of a link. The cost is a function of that link speed. We have something called the reference bandwidth and we divide it by the interfaces bandwidth and that gives us a cost value. Now, on Cisco routers, the default reference bandwidth is 100 megabits per second. So let’s do some math. What would be the cost of that bottom link between R One and R Three? Well, it would be 100 meg, our reference bandwidth divided by ten meg.
The link speed 100 divided by ten, that’s ten. We’ve got a cost of ten going from R One to R Three. What about those 100 meg links? Well, that’s going to be the reference bandwidth of 100 meg divided by the link speed of 100 meg for a cost of one. So the cost is one between r one and r two. Between r two and r three. And as we exit R Three to go down to PC Two, that’s another cost of one because that’s another 100 meg link. So let’s do some comparison. If I went from R one to R two to R three, what would be my cost? Well, it will be one plus one plus one. It’s a cost of one.
One to get from R One to R Two, another cost of one to get from R Two to R Three, and another cost of one to get from R Three out to PC Two. A total of three. What if I took that fewer hop path that Rip would select? If I went directly from R One to R Three, then it would be a cost of ten just to get over to R Three. And then it would be a cost of one to get out of R Three. Going down to PC 210 plus one, that’s eleven. So looking at this clearly, the best path for PC One to use to get to PC Two is to go from R One to R Two to R Three and then out to PC Two. And that’s a look at the theory of open shortest path. First, the OSPF routing protocol.