On Thu, Aug 23, 2007 at 01:29:05PM +0300, Mihai Chelaru wrote:
> David Young wrote:
> 
> > BTW, round-robin is not ordinarily a good multipath routing policy.
> > RADIX_MPATH references gateway "selection by Modulo-N Hash (RFC2991)".
> > Check that out.
> 
> Thanks for pointing me that. The problem with such implementation is that it 
> is protocol depedant.

It has to be.  It can't really work right otherwise.  Other protocols
that care about avoiding path asymmetry when multiple paths between two
nodes in the network are in use define part of the network-layer protocol
header for path selection (see SS7 MTP for an example, and particularly the
treatment of the SLS field by most national variants) but IP doesn't so
you cannot really avoid this issue.

This is one thing that, in my opinion at least, should have been fixed
in IPv6.  But it is a little late now!

-- 
  Thor Lancelot Simon	                                     tls@rek.tjls.com

  "The inconsistency is startling, though admittedly, if consistency is to
   be abandoned or transcended, there is no problem."	      - Noam Chomsky