[j-nsp] IS-IS Badness with JunOS
Mark Tinka
mtinka at globaltransit.net
Mon Apr 6 20:10:52 EDT 2009
On Tuesday 07 April 2009 04:03:39 am Piotr Marecki wrote:
> Are you advertising RSVP LSP as p2p links in ISIS ? ( aka
> forwarding-adjacency feature in IOS ) and/or you are
> using TE shortcuts ?
Neither.
On the Cisco side, we're using the Autoroute feature (tunnel
mpls traffic-eng autoroute announce). We find it's less
complicated than Forwarding Adjacencies, and does the job
quite sweetly.
On the Juniper side, given that we only need to traffic
engineer data toward only three destinations on the other
side, we found pointing static routes into the traffic-
engineered LSP's (lsp-next-hop) to be quite simple and
predictable. We've been lab'ing IGP Shortcuts for the
Juniper side, but we've found we need to have a couple of
trade-offs in other areas, so we're still waiting to
reliably deploy that.
LDP is tunneled within RSVP on both sides.
We've had cases where the circuit between the traffic-
engineered routers fails for regular reasons that circuits
fail, and MPLS LSP's + RSVP signaling along the failed path
is nicely torn down and traffic immediately re-routed to the
backup path (ECMP); and the paths come back up nicely when
service to the failed link is restored - no impact to
customer traffic. So we've not seen any "stuck"
relationship, so to speak, between RSVP and IS-IS before.
The only difference here is we, by hand, restarted RSVP.
> It looks indeed scary, but theoretically FA has big
> potential of doing bad thing...
Indeed.
We're trying to stay away from FA's. They would add
complexity in our case.
> Sorry for not providing answers, but just giving more
> questions :(
Appreciate the input.
JTAC are also checking it out. Will let you know if they
find anything interesting.
Cheers,
Mark.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 835 bytes
Desc: This is a digitally signed message part.
URL: <https://puck.nether.net/pipermail/juniper-nsp/attachments/20090407/53563b50/attachment.bin>
More information about the juniper-nsp
mailing list