Thanks Rick,

I would hate to see the t1 circuit not being used but your point do make a lot of sense. for redundancy to work. Do I need to setup both circuit on one router? Or it can be on different router?

Gerald

For redundancy to work they can be on different routers. The details of how to do it would vary depending on whether you are using a dynamic routing protocol (different protocols would have different implementation details) or using static routes. But for redundancy you want each of the routers with the MPLS circuit to be seen as potential ways to get to the remote destinations and you want the route through the higher bandwidth connection to look more favorable.

I understand your point about not wanting the T1 to be un-used. One alternative to consider would be to implement some Policy Based Routing which would identify some type of traffic (perhaps all FTP traffic, or perhaps traffic from a particular subnet, or some other particular identifiable traffic) and route it specifically over the T1. This would give some use of the T1 without getting into the issues of load balancing.

HTH

Rick

Thanks again Rick, you are giving me a lot of valuable information. I got another newbie question for you. when I do a traceroute from the router that has the t1 to a branch office, it doesn't travel out on the serial connection, instead it goes to the fastethernet and goes to the 9mbs circuit. I believe its because we are running eigrp protocol. Do you have any recommendation on how to make your recommedation work?

Thanks

Gerald

If you are running EIGRP as your dynamic routing protocol then it would logically choose to send all the traffic over the interface with the highest bandwidth. And if it is configured correctly it would choose to use the other interface if the preferred interface became not available.

My suggestion of Policy Based Routing works very well with EIGRP. In essence PBR provides an alternative to the traditional routing decision for certain identified traffic. To implement PBR you would create an access list that identifies the traffic. For example if you decide that you want to send all traffic that is telnet you might create an access list like this:

access-list 123 permit tcp any eq 23 any

access-list 123 permit tcp any any eq 23

Then you create a route map which uses the access list to identify traffic and you set a next hop which will send the identified traffic to a next hop that is different from the normal choice in the routing table. It might look like this:

route-map use-alternate-path permit 10

match ip address 123

set next-hop ip-address w.x.y.z

And then you assign the route map on interfaces where the traffic will be received

interface fastethernet1/1

ip policy route-map use-alternate-path

This link will provide some additional information and should help you get started with this:

http://www.cisco.com/en/US/docs/ios/12_0/qos/configuration/guide/qcpolicy.html

HTH

Rick

Thanks Rick, That really helps me a lot.