I've setup a frame relay network with ISDN routers for their backup internet connection.
Basically, I used HSRP on two 1720's at each remote site. One 1720 is used to connect to the frame, the other is used to connect back using ISDN. If the frame interface goes down, then the ISDN becomes everyone's gateway and dials back to the central site.
The problem is that sometimes the Frame Interface doesn't go down. Just the isp's ability to route through the frame. So HSRP never kicks in and the link is just dead. I've tried using EIGRP, but since the ISDN never dials, the new route is never created.
I've read about Cisco's newer feature Dialer Watch which dials based upon a route's existence in the routing table (while using a dynamic routing protocol). The idea is that if the route over the frame goes away, then the ISDN will kick in and dial.
Does anyone have any experience with this.. and better yet, does anyone have any sample configs using this and HSRP?
Dialer-Watch would work fine if you had the Frame Relay and ISDN links on the same router... and you won't need HSRP. Do you really need router redundancy?
The entire network was sold with a "no single point of failure" promise. There is hardware redundancy in the Clustered windows and unix Servers, Clustered Terminal servers, etc..
I thought I should be able to use both routers and Dialer Watch by using EIGRP on them and then tracking that a route exists on the ISDN router to go back across the frame. When that route doesn't exist, the ISDN dialing would kick in. This would solve my HSRP problem of having the T1 interface in an "up" "up" state, but the frame is actually down.
If there are other ways to do it that don't eliminate the second router I'd love to hear them.
Correct me if I misunderstood what you said... You want to run EIGRP between the routers, through the LAN, and the router with the ISDN link would have as preferred route to some network the path through the router with the FR link... This way if the main router loses the path via the FR interface, the router with the ISDN link would establish the call, because the preferred path has gone...
The problem is that HSRP doesn't kick in because the router with the FR link doesn't lower its HSRP priority because the FR interfaces doesn't go down, right?
Have you tried to use FREEK on the router with the FR link? I'm not saying it will work, it's just an idea...
Yes I want to use EIGRP between the routers. These routers are sharing an address through HSRP, this address is used by local computers as their gateway.
I want to use Dialer Watch (link provided below) to watch the dynamic routing table for a remote route that will only exist if my frame is up.
example: my local network is 192.168.10.x, the remote network is 192.168.20.x
If I am at the remote site and there is no EIGRP route pointing back to 192.168.10.x, then I know that my frame link is down. Dialer Watch (which is monitoring this route in the route table) magically dials from my remote ISDN router and re-establishes the connection.
Birds chirp, children sing, and the client breathes a sigh of relief.
Then when my ISDN router sees that a route exists back across my primary interface (namely through the ethernet interface of my ISDN router, to the Frame router and across the frame) it will shut down the ISDN connection and allow routing to continue normally across the Frame again.
I've attached a link below which describes dialer watch.
On the frame relay interface, use a sub-interface and use the Frame Relay End-To-End Keepalive (FREEK) feature if you have newer code (12.0(3)T I believe) and Cisco's on both sides. When the keepalives are missed then FREEK will force the sub-interface to a down state.
Using HSRP, you could track the sub-interface and when it goes down then the HSRP priority could be adjusted lower then the ISDN routers priority, and using preempt on the ISDN router would make HSRP go active on the ISDN router.
That sounds very cool. I'm already doing the rest, so it looks like all I have to do is implement FREEK.
Thanks, I'll let you know how it works out.
Regarding FREEK, two things:
1- It is a Cisco proprietary feature, so all the routers must be from Cisco
2- You must configure both ends of a VC to send keepalives, so if you don't do it at both ends at almost the same time, I'm afraid one of the sides (the one which will be sending the requests) will consider the PVC down because it doesn't receive replies. You can always increase the event window length to avoid this during the configuration.
Hope this helps.