[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [RRG] BGP Graceful Restart (was Re: Geoff Huston's article ...)
On Jun 20, 2007, at 1:11 PM, John G. Scudder wrote:
On Jun 20, 2007, at 2:39 PM, Tony Li wrote:
The presumption with GR is that a router advertising its restart
capabilities will have a forwarding plane that will continue to
reliably forward during a failure in the control plane. If it
does not do this, then it will either blackhole or misroute
traffic, which is obviously not good.
Note that for GR between IBGP peers, this potential issue is
reduced. That's because a control plane failure which is serious
enough to affect the forwarding plane is likely to also take out
the IGP. Since IBGP routes are typically resolved through the
peer's loopback address which is learned through the IGP, an IGP
failure will have the effect of invalidating dependent IBGP routes,
regardless of their GR status.
Unless, for example, an implementation were to use some different
handling for IP-based stuff (e.g., BGP), versus, say, IS-IS. You might
laugh but I had this occur a couple years back (several times, mind you,
under different conditions, on a very popular router vendor/model) when
IP forwarding fails but IS-IS would stay up, so next hops were reachable
and packets blindly discarded. The interesting bit was that IP
was only failing on certain line cards, but the result was the same.
Of course, the architectural flaw was quickly hacked around, but it was
a reality and I wouldn't be surprised to see something of this sort
recur, in particular when employing non-IP IGPs.
to unsubscribe send a message to firstname.lastname@example.org with the
word 'unsubscribe' in a single line as the message text body.
archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg