[ipv6-wg] IPv6 on P2P links

Inline: GV>

-----Original Message-----
From: ipv6-wg-admin at ripe.net [mailto:ipv6-wg-admin at ripe.net] On Behalf
Of Daniele Orlandi
Sent: 26 May 2011 18:34
To: ipv6-wg at ripe.net
Subject: Re: [ipv6-wg] IPv6 on P2P links

On Thursday 26 May 2011 10:07:39 Jasper Jans wrote:
> Can anyone give me some real world experience with IPv6 numbering on
P2P
> links in their network?

I've not yet been able to figure out another aspect of prefixes longer
than 
/64.

How are TCAMs in hardware routers and routing lookup code going to
handle 
matches longer than 64 bits?

GV> it doesn't really matter as those are destinations that are normally
not in the data-path, but just in the "testing path", so optimization
doesn't really matter for these prefixes for the data-path services.

Are ALL vendors putting all the 128 bits in the TCAMs or is there
someone (or 
most?) with reduced match size that implies a second lookup?

GV> oh.. I think there is more to it then just TCAM... what about
route-convergence preferences etc... However in this case the TCAM is
not a big practical issue I believe.

Or is the CPU-based algorithm going to be slower if the match is greater
than 
the native CPU word size?

That, alone would justify avoiding having prefixes longer than a /64 in
the 
FIB since it would mean reduced PPS throughput for those destinations.

GV> important is that host/user stations are on /64 bit length prefixes,
the point-2-point's 
don't really matter actually in that case as they are just transport
path and not a routing destination.

Is anyone with an insight view of actual lookup method able to enlighten
me a 
bit?


G/