Design help: T1 backing up Metro Enet

If this is set up as independent networks (separate IP address space at each end), it is trivial to implement using basic routing protocols as even RIP would do the job.

If this is set up as an extended LAN, you'll need to enable spanning tree on every link between every bridge/switch that supports a link between the two sites and adjust your metrics so the T1 selected only if no other cross links remain. Again, a trivial setup to implement if you read the manuals and look at how spanning tree does its thing.

A few hints:

• If you're using an extended LAN, you are not going to be happy when fail over to the T1. Performance will not be adequate unless your traffic is minimal (including background broadcasts).

• Whether routing or bridging, you need to routinely test the T1 to detect failure while it is not in use. Otherwise, when you finally do need it, it probably won't be working.

Good luck and have fun!

Vincent -

Thanks for the reply. It would be separate networks. I tried running RIP, but couldn't seem to get it to detect the link as being down. Furthermore, the aging time was atrocious. I was trying to use ping tracking via the SLA Monitor function within my static routes (there are nonly two routes at the distant end since all traffic is routed to the main site). Thsi worked fairly well at the remote site, but my 3750s don't seem to support it (even though it would appear that they should through the Software Advisor). From what I can see, the ping tracking (they call it ICMP Echo Operation) is supported, but it looks like alot of that changed in 12.3 - which isn't available for the 3750s. Would you happen to know if I can accomplish this?

Thanks B

Vincent -

I was just re-reading my original post and realized that I forgot to mention my key problem! The metro ethernet link never goes down in the eyes of my CPE. The transceivers that my providers has given me (Tasman boxes) always have the ethernet link up even if the WAN side is down. I'm struggling with how to actually know that the link failed.

With routers at both ends of the link, any routing protocol should do the job. I'm assuming your metroEthernet goes end to end, not via VPN over the Internet. If the latter, its slightly more difficult (see the white paper on my web site on providing alternative routing for VPNs).

As for response time of RIP, if it otherwise worked, go back to it and adjust the timers to get the response to failure down to something reasonable. You should be able to get failover to function in under 10 seconds whereas with default timers it will take three minutes.

Good luck and have fun!

I'll have to try it again... I wasn't able to get RIP working at all. Let me give it another shot and I'll post back.

Thanks B

One option is to use router tracking objects. Fairly new though, so it'll be 12.3.something. Track ICMP out across, and setup floating routes to handle the case of the MOE being up or down.