High-Level Network Design and Etherchannel Load Balancing

Terry Slattery
Principal Architect

After the experience with a wedged router interface (see prior post: Cisco Router Interface Wedged), several of us at Netcraftsmen were discussing etherchannel load balancing and what to recommend to customers and how to check the setting in network equipment.

James Ventre made a very good point for using src-dst-ip as the load balancing algorithm.  Each pair of src-dst IP addresses will be sent on one of the etherchannels.  His argument is that using IP addresses makes the load balancing operate much like equal cost multipath routing, even though it is a switching algorithm, not a routing algorithm.

Using src-dst-ip allows you to continue to think about configuration and troubleshooting at Layer 3 instead of having to switch between Layer 3 and Layer 2 as you examine and think about the packet flows through the network.

Take James’ idea and apply it throughout the network design.  Where possible, make sure that you are consistent in how your design works and favor design choices that simplify or make packet flows consistent.  What I realized is that by making this simple change in how I look at network designs, I can more easily make choices about alternative configurations.  Using this high-level thought process, one of the configurations will have clear advantages over the others.  It is something that can be documented in network policy documents, which get used in network design, network operations, network troubleshooting, and network management.  That’s a big impact.

Thanks, James, for helping elevate my design thought process.



Re-posted with Permission 

NetCraftsmen would like to acknowledge Infoblox for their permission to re-post this article which originally appeared in the Applied Infrastructure blog under http://www.infoblox.com/en/communities/blogs.html


Leave a Reply