Let's talk routers this chapter. We're going to focus on how routers learn what they know. The whole Internet is based on a series of routers that direct traffic for my system to anywhere in the world. So we want to figure out how it is. Routers know the magic that they know.
One of the main ways that routers know is that they can be told through the creation of manual static routes, which means a network administrator is going to create a series of statements that say, Okay, if traffic is going to be on the 10 network, go this direction.
Obviously, that's not what's happening on the Internet,
but for certain situations. In my internal network that may be something that's used.
Usually, we allow our routers to learn about their network and other networks through a series of dynamic routing protocols. There are a handful, and we'll talk about our I P O S P F and B G P in the next section.
B G P is what the Internet routers are using.
We'll also talk about a future that many routers have called an AP, which is network address translation, and it's best friend p 80 port address translation. And then we'll talk about access control lists, which are how we add security to routers,
prostatic routes. We can use command route. Add. There's also rupee for persistent and route print.
These are commands will want to know for the exam.
When we're creating static routes, we use the command route. Add what we want to configure is how to get the destination network.
If I want to get to the destination network on what interface on the router should traffic be sent? Where is its next hop? It's a little bit challenging or not so much challenging as it is time consuming to go through and configure static routes on all your routers.
However, if I do have a specific pathway that doesn't change frequently or doesn't change at all, it might be good to use a static route. In that instance, because dynamic routing protocols are kind of chatty,
they generate some additional traffic on the network. If you have a route that doesn't need to be learned or isn't something that comes and goes in a static route is probably best
for many of the roots, you need your devices to learn a routing protocol is probably the way to go.
There are three main ways of routing protocols. One is called a distance vector. Then we have Link State, and then we have our exterior gateway protocols,
the ones you and I are most likely to work with our distance factor or link state
for distance vector routing. These routers make their decision on the best path, based on how many hops away and in what direction. For instance, a distance vector router is always going to pervert two hops to three hops, even if the three hot direction or three hop link is at a much faster speed
distance Vector routers are older than ***, which is routing information protocol.
We're on rip version two, but it's still fairly dated. The thing about rip and distance vector routing protocols is that they're very easy to set up, and they don't generate a lot of traffic on the network.
They're pretty manageable for a very small network, quick and easy we like quick and easy.
The way these routers learn from the network is from their neighbor, and their neighbor learns from their neighbor. That's a method called routing by rumor and the illustration you can see Router C is connected to the four network into the three. Network B is connected to the three network and the two network
A is connected. The two network and the one network
each router only knows about its directly connected network. See only knows how to get to the three network, and the four network be only knows how to get to the two and three. He only knows how to get to the one and two.
Ultimately, you can see that, See says, Hey, I have this information and it's passing its routing table and to be since its routing table into A Until they get to updates about the various networks that exist, that's routing by rumor. The problem with that is that it takes a long time for what we refer to as convergence, which means learning if there's a network change
Another problem is we see that Router C is directly connected to the forward network because B is one hop from See if you look at bees routing table and what's highlighted in green, it says okay and one hop away from foreign network a says. Well, I'm two hops away from the foreign network because I'm one hop from B who's one hot from C.
That's how routing by rumor is supposed to work.
Let's say that that link between B and C goes down be can no longer get to Router C and can get to the fourth network.
So be starts to listen to A and A says I'm two hops away from the four Network B says Well, I'm a hop away from a I must be three hops, see listens and says I must be four hops and then be here, Seen says. I must be five hops and a says I must be six hops.
Basically, what's happening is a process called Count Infinity.
It winds up happening if there are no controls in place to keep these routers from learning from each other, going different pathways when there's a down link.
One of the things that kind of cracks me up is the way that ripped deals with that count Infinity issue.
It says Infinity, as the number 16 16 is a high as we go to infinity.
What that means is once we count to the infinity process, be says, I'm 15, hops away and he says, Oh, that makes me an infinite distance. Therefore, there is a down linked.
That's one of the ways we deal with Count to Infinity, and that's called poisoning a route.
The other thing about poisoning a route is poisoned. Rivers basically means that once that route is down, see sends a message out saying 15 hops away from network for and then being a automatically know that's a dead network.
That's really how we counteract count to infinity.
There's also something called Split Horizon, which means that if you look at counter see, it's sending information about the foreign network on its Ethernet zero interface.
What split horizon means it's not going to learn information about the four network on. It's either Net zero interface.
It's not going to send information to be an A and learn information from DNA about the specific link.
Basically, these are just controls that are built into our I P that can be enabled to deal with that problem called Count to Infinity
with link state routing protocols. The most common is called O S P F
Open, shortest path first.
Nice thing about Oh, SPF is. Rather than learning about the network from neighbors, oh, SPF allows each router to learn about the network on its own and create its own topology table.
Every router directly communicates with every other router through what are called Elsa's or link state advertisements.
It continues to make sure that the network is available through these little link state advertisements because they're essentially constantly checking in to make sure that they're still there.
When that router doesn't hear the L S. A. From a specific router, it assumes that it marks that router is being dead or not available and then learns different pathways throughout the network. The problem with this is that it's very resource intensive.
Rip is good for smaller outer routers that don't have a lot of processing capability, but with link state routers or link state protocols, you don't have a lot of memory, and that's a lot of processing capability to build these topology tables. To track this information,
it's very resource intensive and is designed for larger networks.
Then you have border gateway protocol, and we really aren't going to get into the border. Gateway routers because these are the routers that are used throughout the Internet.
That's a discussion for another day