Re: Unicast RPF path

[Polly Huang <huang@catarina.usc.edu>]

It depends on what's the rest of the topology. If there are multiple
routes between 1-5 with equal costs, you may have asymmetric routes.  For
example, assuming equal-cost links,

	1---2---3---4---5
        |               |
        +---6---7---8---+

1 might choose 2 as the next hop to 5, while 5 might choose 8 as the next
hop to 1.

cheers,
-Polly

On Fri, 5 Feb 1999, Shuqian YAN wrote:

> 
> could someone help to answer the question below...
> 
> thank you,
> 
> shuqian
> On Fri, 5 Feb 1999, Shuqian Yan wrote:
> 
> > Hello, all
> > 
> > This is a question about unicast RPF model used in ns2.
> > 
> > If I have the following partial topology,
> > 
> > 1-----2-----3------4------5
> > 
> > numbers are representing nodes and the links between nodes are duplex
> > links.
> > 
> > if some agent in node 1 unicast a pkt to the final destination,
> > which is node 5. since I also want the pkt to be received by the 
> > intermediate node, in my algorithm, I always use rpf-nbr to find 
> > the next hop where the agent will be connected, the pkt is thus
> > forwarded hop-by-hop based on RPF until finally reached node 5. 
> > The overall path in this case is 1->2->3->4->5.
> > 
> > suppose node 5 unicast a pkt toward finaldst node 1, and again I want
> > the pkt to be received at intermediate nodes, so I also use the rpf-nbr to
> > find the next hop towards node 1. My question is: is the path for node 5
> > unicast pkt hop-by-hop towards node 1 will be the reverse path of node 1
> > unicast pkt hop-by-hop towards node 5, or is the path will be
> > 5->4->3->2->1?  could the path be different for the two cases I described
> > above?
> > 
> > thank you,
> > 
> > shuqian
> > 
> > 
>