« previous entry | next entry »
Mar. 13th, 2006 | 03:52 pm
posted by: null_variable in cptnotes
To recap, we talked about routing protocols (protocols routers use between themselves).
Routed protocols are what actually travel through the router.
The dataflow through a network when you’re coming from a PC, going through a few routers, etc, generally on the two PCs the data’s going to go from the application layer to the physical layer, but it won’t get up to the application layer during the intermediate spots.
If I’m sending data from PC1 to PC2, the application will need access to the network. So they’ll move down the layers from PC1. What’s traveling from the PC to the first router is a frame, and in this case, an Ethernet frame. The Ethernet has a start up frame delimiter…the frame’ll also have a destination mac address of the router and the source mac address of the pc. (Also, the router can detect if there are errors in the frame by checking the cyclic redundancy check (CRC) [at layer two]. If the frame is corrupted, it’ll re-request it from the PC. Then you only go up two layers, rather than going the whole way up.) If there are no errors, the router strips the frame, and at layer 3 it looks at the destination IP, and once it realizes it’s not for itself, it looks into its routing table and try and figure out where to send it next. So the destination mac address is the router, but the destination IP address is for PC 2. So the router decides to send it to R2, then it drops back down to layer 2 (connected with an X.25 link). R1 takes the packet, an d since it’s directly connected to the x.25 link, it knows it has to take the packet and put it into an x.25 frame (which is very much like an Ethernet frame, made for a different protocol). It puts it into a frame, and between R1 and R2 in layer two, the source and destination address have changed. The frame goes into R2, and it does the same thing with the protocol (frame comes in, goes up to layer 2, looks at the frame, error checks, goes to layer 3 to check destination/source ip addresses, then goes back to layer 1). So routers, unless they’re fancypants, you’re only going to layer 3. Routers work at level 3.
Page 426-7 – for your own “enjoyment”
When a frame comes into either a bridge or a switch, it looks at the destination mac address and decides what port to put it out, to flood it, etc. It switches criteria on a destination mac address. With a router, it has a frame come into it, looks at the CRC, then tosses the frame and looks at the destination ip address, looks at the routing table and decides what interface to send it out of – you can think of this as switching; from one interface to another. However, it’s making forwarding decisions, not switching decisions. So routing is forwarding.
Page 444 – routing protocols. 444-446. They talk about some that you will see time and time again. RIP V1 & RIP V2. Most small networks will run RIP. The difference is V1 can’t handle classless subnetting.
Classless routing – no class a, b, c. What that means is I can have ie 10.1.16.3 with a subnet mask of 255.255.255.248. With classless subnetting, you could also have 192.168.50.4 – class C address – but having a 255.224.0.0 as a subnet mask – a class A address. Classless interdomain routing (CIDR) is a way for ISP’s to allow routers to not have as many routes. Variable length subnet masks (VLSM) lets you have different subnet masks on the same router. These all fall under the guise of classless subnets.
RIP is a Distance Vector Protocol. DVPs are typically simpler than other styles of routing protocols.
IGRP is another DVP very much by RIP, developed by CISCO to fix RIPS problems (RIP uses a hop count). IGRP’ll give you 11 hops default. It also gives you a triggered update. With RIP, if you lose one of your connections, your router won’t tell anybody until 30 seconds from now (when it broadcasts any 30 seconds). IGRP tells everyone as soon as the connection is lost. It also allows you to have more than one path to a network.
EIGRP – almost identical to IGRP, except that it’s a hybrid routing protocol (the two families are interior that are distance vector or interior that are link state).
OSPF – this reports on the state of the routers interfaces. It’s a link state protocol. It’s very quick to converge (it’s very quick in a link state network for all the routers to have all the same information, because they all know the whole topology of the network). There are different implementations of OSPF. It’s very common. It’s more complex in the fact that it takes more processing power off your router.
ISIS is an OSI routing protocol that actually works kinda well.
BGP is a border gateway protocol, worked on networks much bigger than on the campus. It’s used on the internet.
What you need to know for a test essentially is what’s on page 444 – 446. You should be able to answer questions based on those pages.