Type one and Type Two LSA's

DevilWAH

Can some one tell me if I have this right please.

In a multi-access network

All routers send Type 1 LSA's to the DR only (not to other non DR routers)

the DR floods the type one LSA's by way of a Type 2 LSA to all other routers

So Type 1 from others ----> DR

Type 2 from DR to ----> others


In the ospf data base it seem all routers have Type 1 LSA's from all other routers, are they exchanging these directly or through the DR??

Cheers

Aaron

DevilWAH

OK no, been reading around more and it seems all that the Type two does is list every router on the link.

Just trying to work out why this is needed, seems a bit redundent.

Ryan82

Found a pretty good link that describes it: https://supportforums.cisco.com/thread/2104610

Forsaken_GA

DevilWAH wrote: »

OK no, been reading around more and it seems all that the Type two does is list every router on the link.

Just trying to work out why this is needed, seems a bit redundent.

They're kinda the same information, but serve different purposes.

Ok, remember that on a multiaccess segment, the routers do not form an adjacency with every other router on the link. They only establish an adjacency with the DR and BDR. So the routers generate their Type 1 LSA's and send them up to the DR/BDR. Non DR/BDR routers won't talk to each other on a multiaccess segment, beyond establishing to the 2WAY state, so they don't know the routes each other has.

This is where the type 2 LSA comes in. The DR/BDR basically aggregates all the Type 1 LSA's it receives, and sends that data back out to all of it's neighbors in Type 2 LSA's.

Edit: The link Ryan provided is good, and addresses the scalability issues that would be involved with a full mesh of neighbors on a multiaccess network.

DevilWAH

See that is what I though, but acourding to this OSPF LSA Details

the Type 2 simple containers a list of the routers on the network segment

Type 1's containt the network information about the network, and this is reflected if you capture the packets, the DR floods Type 1 back out, not type 2. It when it gets a LSU from a router on the nestwork with type 1 LSA's it floods them back out as still type 1's.

The LSU now has the DR ID, but the LSA's it contains still have the originating router ID

So from that you can build up the network even though only the DR and talks directly to others. Because the LSA are flooded its possible to rebuild the network topology from them.

But now I think I get it, by defining the multi-access network and who is a member of it, the indvidual routers don't need to advertise this information. So by the DR defining the network, all the other routers can remove his information from there LSA's. So it is not an summary of all the LSA's, connected network are still advertised by the other routers to the DR and re flooded as type ones. The Type two jsut summarises the topology of the multi-access part of the area.

DevilWAH

So in the case of having three routes on a network you would need LSA's that say

R1 --> R2 use 192.168.1.0/24 network interface
R1 --> R3 use 192.168.1.0/24 network interface
R2 --> R1 use 192.168.1.0/24 network interface
R2 --> R3 use 192.168.1.0/24 network interface
R3 --> R1 use 192.168.1.0/24 network interface
R3 --> R2 use 192.168.1.0/24 network interface

So each router would advertise it can get to all other routers as a indivual LSA.

The DR simple advertises R1, and R2 and R3 are on the 192.168.0.0/24 network by use of a type 2. cutting down on the number of LSA it floods.

however if R2 had a connection to 172.16.23.0 / 16

it would still create a type 1 LSA for this, and forward it to the DR by multicast. he Dr would then flood it out to the rest of the routers on the segment using a type 1 LSA.

Correct me if I am wrong but that's my take on it

Forsaken_GA

Ok, yeah, I had my head wrong. Had to go to the RFC to get a handle on this.

Near as I can tell, the point of the type 2 LSA is to allow the multiaccess network to be represented as an individual node within the SPF calculation. Instead of it having to run based on every single router in the network, the type 2 LSA allows the SPF algorithim to essentially short circuit and consider that multiaccess transit network as a single node

DevilWAH

Forsaken_GA wrote: »

Ok, yeah, I had my head wrong. Had to go to the RFC to get a handle on this.

Near as I can tell, the point of the type 2 LSA is to allow the multiaccess network to be represented as an individual node within the SPF calculation. Instead of it having to run based on every single router in the network, the type 2 LSA allows the SPF algorithim to essentially short circuit and consider that multiaccess transit network as a single node

I love the way that on one of the CISCO pages I was looking at said "Type 2 are the simplest to understand.." and then went on with one line to describe them. I was like What???

Now I have run a few traces and seen it in action it starts to make sense and 'I see why you have them, exactly like you say wrap up the multi access network as a single object in the database. I found if you add about routers to your topology you start seeing it in action, with only 2 or 3 its easy to miss its purpose and how it works.

Cheers for the help guys

vinbuck

I had quite a time getting my head around LSA types for ROUTE, in the end though, I just kept changing ospf network types in my lab to see what effect it had on the OSPF database. If you explore the OSPF database and look at the data being provided in each entry, then you will start to match up the theory to the practical application.

It's also really helpful to be familiar with the OSPF database when you start doing heavy route filtering on OSPF