Data Path Diversity

[r.amoranto]

I have a 2611 running EIGRP using 2 frame-relay T1s that are going to the AT&T frame/ATM cloud. That router has several sub-interfaces with DLCIs going to different parts of the country within our private network. My mission is to create redundancy and to have the data go to a totally diverse path (CO) in case both T-1s fail. I will be introducing another router with another frame T-1, and will use HSRP. How do I do this?

[thisisshanky]

What you can do is configure the router with 2 T1s as ACTive, and the other one as standby. Now when a T1 subinterface fails, you can track this failure using standby track command and switch the active router to the second router. But this would create issues, as follows.

If one pvc goes down,can result in shifting the whole routing (ACTIVE HSRP router) to the second router. So this wont be a wise thing to do.

You might want to try this instead. For each site, order two PVCs. A higher CIR pvc for your primary link from each site to HQ, while a lower CIR pvc to the new router that you are adding. EIGRP will converge over the higher CIR link. (you can use bandwidth statements to manipulate this). When this pvc fails, the site will be still up on the lower CIR pvc.

You can still run HSRP on the ethernet segment of both the routers. You dont need to track the serial interface failures. If any remote site goes down, route would converge through the new router. So any packet going out from HQ to the remote site (that is down) would first go to the active router first. Since the pvc is down there, and routes for that site will now converge through the new router. Now the active router will forward packets to the new router and packets would go out to the remote site.

[steve.barlow]

HSRP will be for the LAN side of the 2611 and the other router (if using static routes or have PCs pointing to them - if not, EIGRP can take care of things for you). This link can help: http://www.cisco.com/univercd/cc/td/doc/cisintwk/ics/cs009.htm .

As for the WAN side, you can to add a second (or third, not sure how you have set them up) PVC (DLCI) at each remote site that will terminate on the new router's T1. This way, the remote site will EIGRP peer with both routers, and if one end drops, it can still send/receive from the remaining peer. Depending on the new PVCs bandwidth, you can load balance across the two PVCs - equally or you can use unequal load balancing with the variance command. The remote sites serial WAN interface will have subinterfaces for each central site router.

As for diverse paths, that must be worked out with the Telco but is a very good idea.

I have done something simliar once before and this is a sample from the remote side:

interface Serial0/0

description Connection to Toronto RouterB & RouterA - S0/1

backup delay 5 10

backup interface BRI0/0

ip address 10.1.2.1 255.255.255.0

ip directed-broadcast

encapsulation frame-relay IETF

no ip mroute-cache

service-module t1 timeslots 1-8

frame-relay map ip 10.1.2.2 50 broadcast

frame-relay map ip 10.1.2.6 62 broadcast

frame-relay lmi-type ansi

!

interface Serial0/0.75 point-to-point

description Connection to Toronto DLCI 75 - routerA - S3/1

ip address 10.1.1.1 255.255.255.0

no ip directed-broadcast

no cdp enable

frame-relay interface-dlci 75

!

This is a bit messy but it should give you an idea. You can do yours with two subinterfaces, one pointing to each router.

Hope this is clear, if not let me know.

Steve