Link State feature

[vishalpatil86]

hi all,

I have a question on link state routing protocol feature

One of the features of link state routing protocol is updates are multicasted                  

And other is updates are sent directly to neighbors. Then what is the use of updates being multicasted?

[Richard Burts]

Vishal

I do not believe that it is correct to say that a feature of link state protocols is that updates are multicast.

What is your source that says that a feature of link state protocols is that updates are multicast?

By definition neighbors are formed with other OSPF routers that are directly connected on the same subnet. The use of multicast is that the DR sends a single update packet to the multicast address and all the neighbors on that subnet receive it. The alternative would be for the DR to send unicast updates to each individual neighbor.  Think about a situation where there is a subnet with 10 OSPF routers. When the DR needs to send an update is it more efficient to send one multicast packet (which will be received by 9 neighbors) or to send 9 unicast updates, one to each neighbor?

HTH

Rick

[vishalpatil86]

so, the updates are unicast, mulricast or broadcast?

[Richard Burts]

Vishal

If you ask your question in terms of a specific routing protocol, such as OSPF or IS-IS then we can answer how the updates are sent. But when you ask the question in terms of link state protocols in general there is no answer because there is not anything in the link state protocol that specifies the update method and it becomes an implementation detail for the specific protocol.

The important characteristic of link state protocols is that they flood link state updates through the routing domain and then use the link state updates to create a map of the topology and from the topology table they then choose the shortest path to destinations which become entries in the routing table. There is not anything in the link state protocol that specifies how updates will be sent.

HTH

Rick

[vishalpatil86]

Richard,

Thank u for the reply