Re: Flow label versus Extension header - protocol itself

[Iljitsch van Beijnum <iljitsch@muada.com>]

On 9-mei-2005, at 19:45, Erik Nordmark wrote:

> > Simply going for the option means adding bytes to all packets   
> > forever.

> Nobody has argued that there would be extra bytes in very packet.

> We only need to carry the demux information to the peer when the  
> locator pair is different than the ULID pair.

I know, and I didn't mean "all" packets, but rather all packets that  
are actively shimmed. Currently that may not look likethat would be  
too many, but if we want to rehome using the shim for traffic  
engineering purposes, or use unreachable ULIDs, this could be a lot.

So a moderate investment in future optimization is prudent, IMO.