[nsp] High-availabity GigE connection to two switches

[Robert A. Hayden]

The reason you want a routing protocol is for seamless failover.  If you 
have two links and you weight one default heavier than the other, you 
might have a partially failed link but not a failed layer-1 connection.  
Relying on a routing protocol from some upstream router to provide your 
default addresses all of your L1-3 issues in one bite.

In some cases, even the failure of the link might not yank a static 
default route out of the server's table.

If you are comfortable with the possibility of doing some kind of manual 
inspection to address a failure, then there are non-routed ways of doing 
it for sure.  I just know that the way I proposed, while older than dirt, 
works pretty good for most failure modes.

- Robert

On Tue, 1 Jun 2004, Hudson Delbert J Contr 61 CS/SCBN wrote:

> 
> Tony,
> 
> 	first of all, i don't think you're weird as simple is better if no
> other
> 	issues are prevalent.
> 
> 	I hesistated to reply as i have not seen the entire thread but
> tony's is the
> 	approach i would take.
> 
> 	if not needed, why run any routing protocols at all if this server
> has static
> 	routes of volume small enough to handle by hand then why run more
> code.
> 
> 	there are questions i'd like to ask about what the network looks
> like
> 	that needs the server.
> 
> 	show me the topology and/or describe why you need to run gated or
> routed
> 	at all
> 
> 
> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net
> [mailto:cisco-nsp-bounces at puck.nether.net]On Behalf Of Robert A. Hayden
> Sent: Tuesday, June 01, 2004 1:27 PM
> To: Tony Li
> Cc: Rubens Kuhl Jr.; cisco-nsp at puck.nether.net
> Subject: Re: [nsp] High-availabity GigE connection to two switches
> 
> 
> Yea.  Just have RIP or your protocol of choice announce a default and have
> gated receive it on the server.  If a link goes down, magic occurs and you
> keep going.
> 
> On Tue, 1 Jun 2004, Tony Li wrote:
> 
> > >> Any suggestions on how to provide high-availability to a server that
> > >> has two network cards, each one connected to a different switch ?
> > >> Etherchannel doesn't seem to apply here, unfortunately. (Switch is
> > >> either a Cat4K or a Cat6k)
> > 
> > 
> > My preference would be to take the obvious approach and make them two 
> > separate
> > subnets.  But then I'm weird...
> > 
> > Tony
> > 
> > _______________________________________________
> > 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/
> > 
> 
> _______________________________________________
> 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/
> _______________________________________________
> 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/
>