[Csci551-talk] Need a few clarifications...

[John Heidemann]

On Sun, 22 Feb 2004 18:08:11 PST, Nirav Jasapara wrote: 
>
>--- Devarshi Shah <devarshi at ieee.org> wrote:
>
>> a) The project definition says that "... Note that
>> each router will need
>> to send multiple, increasingly longer routing
>> messages....". How would
>> this happen ? I mean, in Link State routing, I
>> thought that the 
>> contents of the LSA message were the routes that the
>> originator is
>> responsible for, and the routers that the originator
>> is connected to.
>> This does not change when you reach steady state
>> (i.e. having discovered
>> your neighbours) unless routers go down... And so,
>> shouldnt the length
>> of the message be fixed (once you have discovered
>> who your immediate
>> neighbours are) ?
>
>'increasing long routing messages ' refers to the
>startup stage when the routers start building their
>routing tables and sending out LSA messages. After
>some time this should converge and then steady stage
>is reached.

Actually, not quite... my statement in the proposal about increasing
message sizes was simply incorrect.

>
>> b) Is a Router supposed to forward a LSA message
>> that it receives from
>> its neighbour to the other neighbours ? (Apart from
>> sending out its own
>> LSA message)
>
>a router will send a LSA message only when it observes
>some change in its routing table. I dont think there
>is any _forwarding_ done.

That's not true for LS routing (although it would be true for DV).

In LS routing you send out messages when your LINK status changes,
and nodes propgate these via reliable flooding.

   -John Heidemann