[c-nsp] Failover with multiple upstreams running BGP

[Rodney Dunn]

That sounds a bit long.

If you stop advertising the prefix and monitor a looking glass
how long does it take to go away there?

There isn't anything you can do if the problem is route
propagation delay upstream.

Rodney


On Tue, Feb 19, 2008 at 05:52:24PM +0500, Junaid wrote:
> Hi,
> 
> I need advice on how to have a fast converging failover from my
> upstream Internet providers. I am an ISP with three Internet upstream
> providers. I am running BGP with them. Currently I am load sharing my
> Internet traffic by advertising some of my /24s via one ISP and some
> via the others. However, as a backup, I advertise the supernet(s) via
> all links so that in case one of the links go down, traffic should
> return via the supernet. But it seems that after one /24 advertisement
> stops reaching the Internet, traffic of that prefix comes via the
> supernet advertisement after quite a while (even upto 30 mins). What
> is the reason for this behavior and how can I design for a faster
> convergence in such a case (i.e. one of my upstream link goes down)?
> 
> Thanks.
> 
> Regards,
> 
> Junaid
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/