Dual ISP Question

Is there a way to have a router with an ethernet connection to two ISP's (and an inside interface of course) fail over from one ISP to the other when traffic stops flowing on the first one?

• Neither ISP shares it's routing table. (it's not an option so don't suggest it)

• Neither interface that faces the ISP's will go down (both ethernets attached to a switch)

I was thinking of something with a route-map to a known "pingable" address - force it out primary ISP interface - but how do you make it fail-over to the backup ISP.

Of course I could write a script that runs in conjunction with the route-map, but I'd really like to keep it on the router.

Suggestions??

___ ___ ___ ___ / \\/ \\ / \\/ \\ | ISP A | | ISP B | \\___/\\___/ \\___/\\___/ =A6R=A6 =A6R=A6 +-+ +-+ =A6 =A6 =A6 +--------+ =A6 +---=A6myRouter=A6---+ +--------+ =A6 =A6 +--------------+

Some platforms with some versions of IOS support ping-based detection of route availability. I do not recall any of the details; probably one of Vincent Jones' messages names the technology.

Which platform and version are you working with?

Couldn't you just set up 2 default routes, 1 with a higher administrative distance than the other?

Say,

ip route 0.0.0.0 0.0.0.0 dialer1 ip route 0.0.0.0 0.0.0.0 dialer2 220

I th> > >* Neither ISP shares it's routing table. (it's not an option so don't > >suggest it)

the feature you want to use is known as

"Reliable Static Routing Backup Using Object Tracking"

The IP addresses you ping should be somewhere on the ISP's core links, not just on the ISP's POP routers as the POP may become isolated from the rest of the ISP network.

The weighted routes wouldn't work as interface won't be going down to remove the static route.

This looks like what I'm looking for. - Thanks. The command have changed slightly from 12.3(8)T to 12.4

The command "rtr" has changed to "ip sla monitor"

I'll try to post an example config when I have it.

Thanks again J.Cottingim

When I last tried this (a couple of years ago) there were two major gotchas you would need to work around, a search for "ping based routing" should get you more details:

1 - NAT translates in effect would not go away, and policy NAT was only applied when a new translation was created. Only a problem if the router is doing NAT as well as routing. 2 - Under some conditions (never fully quantified) the response time recorder would stop probing a down link, preventing fail back to the preferred link.

Number one could be worked around with a hack, but I never saw number two in a repeatable lab setup allowing full characterization, so I don't know if it is an avoidable feature or a bug.

Good luck, have fun, and report back :-)