ietf
[Top] [All Lists]

Re: [mpls] [tsvwg] [lisp] draft-ietf-mpls-in-udp was RE: gre-in-udp draft (was: RE: Milestones changed for tsvwg WG)

2014-01-21 10:02:56
On (2014-01-17 15:28 -0500), Curtis Villamizar wrote:

This sounds like a router acting as a host for the given packet.  The
forwarding data paths in routers are usually protected.  If the router

This was transit packets ingressing as IP packet, egressing as MPLS packet.
Consequently the peer router needed to touch TTL and checksum. Then generate
new L2 frame.
MPLS core obviously need not to worry about packet being broken, as frame is
fine.

was forwarding and mangled packets, that speak well for the router
vendor since the major vendors claim to protect their data paths from

Are such claims realistic or practical? I think it sounds like snakeoil when
someone claims such absolutes.

this sort of thing.  Less so for the path from the route processor.

This was forwarding plane.

You've made a good anecdotal case for keeping TCP and UDP checksums.

This would actually require IP checksum specifically, otherwise we would have
probably never found it. If each egress PE suffers from it maybe once 30min,
each customer suffers from it less than once a month, even if everyone would
catch it and open trouble ticket, we couldn't correlate the cases to any
single peering router, without having historic routing data for each egress
PE.

Disclaimer, I'm not for or against checksum/FCS in any particular layer, I've
not thought about the matter deeply. It was just extremely timely thing in
real-life network I wanted to share.

-- 
  ++ytti

<Prev in Thread] Current Thread [Next in Thread>