[afnog] Backup routing Engine
mark.tinka at seacom.mu
Tue Jun 12 14:35:47 UTC 2012
On Tuesday, June 12, 2012 03:59:20 PM ibtisam jamal wrote:
> > Thanks for the link but what i wanted was to have both
> > Routing engines operational for redundancy.
> > one as master and another as backup.
> > In case the master faces issues a graceful switchover
> > happens.
What you're looking for is known as GRES (Graceful Routing
It's also well-documented:
Please note that GRES only enables redundancy for the
routing engines. It does not do the same for protocols.
If you want protocol redundancy, e.g., for IS-IS, OSPF, BGP,
RSVP, LDP, PIM, RIP, e.t.c., you'll need to look at 2x
1. IETF Graceful Restart.
2. NSR (Non-Stop Routing).
IETF Grace Restart is now quite mature for all the protocols
I mentioned above, and more. But it requires the devices on
the other side of the link support it too. It's also an open
standard, so if different vendors implement it, it "should"
work between them.
NSR is locally-significant to the router. Unlike Graceful
Restart, NSR simply copies protocol state from the primary
routing engine to the backup one. In case of failure of the
primary RE, the backup one assumes control using protocol
state it had the time of failure of the primary RE. In
theory, this should minimize protocol convergence time
during an RE switchover.
However, the only issue with NSR is that there is a higher
potential for an NSR-related bug taking down the primary RE
and propagating that issue to the backup RE, rendering the
whole solution moot. But it's just a risk :-).
Also, NSR support for various protocols tends to happen much
more slowly than for the Graceful Restart, so you'll end up
needing to run bleeding-edge code if you're an NSR fanatic
and want all your protocols to be NSR-aware.
You can read all about the solutions here:
Hope that helps.
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 836 bytes
Desc: This is a digitally signed message part.
More information about the afnog