Introduction

Using default EIGRP setup, it's impossible to achieve desired load-sharing between locations Chicago and San Francisco in the network shown below:

Explanation of the problem

Chicago router has two equal-cost routes toward San Francisco LAN, as the cost of these routes is dictated by the lowest bandwidth on the path - 64 kbps between San Francisco and San Jose. The traffic going from Chicago toward San Francisco will be equally distributed on both links between Chicago and San Jose, whereas the desired ratio would be 2:1 based on the link speeds.

Exercise

How could you modify the EIGRP design in network shown above to ensure proportional load balancing from Chicago toward all destinations in the San Jose area? 

Solution

There are three solutions to this exercise:

• modify the bandwidth on the San Francisco - San Jose link to be higher than 2 Mbps. The cost of the routes between Chicago and San Francisco will be dictated by the new minimum bandwidth on the links between Chicago and San Jose. This solution is not recommended as it may lead to WAN-related problems due to misconfigured interface bandwidth.
• Run another routing protocol (for example, RIP) between San Jose and remote sites in Bay area and redistribute routes toward the remote sites into the EIGRP process with a high bandwidth, for example using command

router eigrp 100
 redistribute rip 5000000 10000 100 100 1500

• Run another routing protocol (for example, RIP) between San Jose and remote sites in Bay area. Deploy BGP in the network and redistribute RIP routes into BGP. The routes of remote sites in Bay area (for example, San Francisco) are no longer in EIGRP topology table, so the routing from Chicago toward sites in Bay area is influenced only by the EIGRP cost of BGP next hop (San Jose router), resulting in desired load sharing. Please refer to Case Study #4 in Chapter 9, "Integrating EIGRP with other enterprise routing protocols" for more details.