Re: [ns] static route path!

[Lloyd Wood <l.wood@eim.surrey.ac.uk>]

On Sun, 18 Feb 2001, Sowmya Manjanatha wrote:

>                            ----- node2 -----
>                           |                 |
>       src_node --- node1 -                   -- node4 --- dst_node
>                           |                 |
> 	                   ----- node3 -----
> 
>       Given the above topology and considering that no dynamic routing
> mechanism is used, I presume that when the agent on src_node and agent on
> dst_node  are connected using the "$ns connect $src_agent $dst_agent"  command,
> the route between node1 and node4 is setup (correct me if I am wrong).
> 
> 1. If "connect" is not the command when the path is defined, please let me
>    know when it is actually defined.

when the routing table(s) is/are built. Since ns builds datagram
networks, there's no explicit hard-state virtual-circuit path per se.


> 2. In the above scenario, then what's the route chosen, through node2 or
>    node3 ??

depends. If you're setting link costs with DV routing, you can control
the choice, or with multiPath_ even use both paths. Without that it's
likely to depend on how the global/individual routing tables get built
and the ordering within them, and may even come down to your ordering
of node and link creation statements in your scripts.

> 3. Also, is there a linked list for each route in the network which
>    maintains the list of nodes involved in forwarding ??

Er, no. For large networks that would scale much worse than routing
tables defining next hops to destinations. Routing tables are per-node
(or even shared between nodes as a simulation speedup if you're not
interested in routing transients), rather than per-path.

L.

<L.Wood@surrey.ac.uk>PGP<http://www.ee.surrey.ac.uk/Personal/L.Wood/>