Policy Based Routing

[ayokunles]

Hi all,

i need help with Policy based routing.    The branch router has two connections via GRE tunnel to HQ. all routers in question using EIGRP. I want all finance server traffic to use one of the connections to HQ and all other traffic to use the other connection. I am a bit scared of using this because if the finance traffic link fails, would it automatically use the other link and vice versa.  guys, i need your help here.

thanks

[yashfaqu]

Hi,

You can do it with EIGRP, just increase the metrics in a way:

Primary : Finance server routes --> Lower

              Other routes --> higher

Secondary : Finance Server routes --> Higher

                 Other routes --> lower

So in this way, your finance server traffic will go through Primary, and rest will go through Secondary, and if any of the link will go down, the other will provide the backup.

I hope it helps, if i understood your problem correctly.

Yasir

[ayokunles]

thanks Yasir,

i was hoping to achieve this using the route-map command. Would it be possible using the route-map command and still achieve redundancy?

[ayokunles]

Guys still need help

[Richard Burts]

The details of how it would work depend on some things about how your network is set up that we do not know. In general it is quite possible to use a route map in Policy Based Routing to send traffic of one server over one tunnel and all other traffic over the other tunnel.

The configuration of redundancy gets a bit more complicated. Your route map will probably use a set next-hop command to direct the server traffic over the interface. If the next hop specified is on a directly connected interface and if a problem in the connectivity will bring the interface down, then the server traffic will not use the set next-hop address and will fail over to the normal routing table. This happens if the next hop is over a traditional point to point serial interface. But if the connection is over an Ethernet interface then it is not dependable, because many connectivity problems are on the remote end but the Ethernet will stay up. There is an option in PBR to verify reachability which uses IP SLA or Object Tracking to make sure that the next hop is reachable. Using this option should allow your PBR to also provide fail over redundancy.

HTH

Rick

[ayokunles]

thanks rich,

can the routemap statements be confgured on both sides

[Richard Burts]

Yes the route map statements can be configured on both sides. And usually PBR would be configured on both sides to have the traffic be symmetric.

HTH

Rick