LACP with Cisco and VDS - loadbalancing mode does not match?!

[Thomas_]

Hi guys,

I have a VDS (5.1) setup with LACP and load-balance by IP hash with 4 NICs.

On the physical side I have two Cisco 3750 in a Stack. Everything is regarding LACP/port-channel is configured.

When I do the health checks in the Webclient everything is fine (VLAN, MTU and Teaming mode).

However, when I run show etherchannel load-balance on the Cisco I get src-mac.

I think it should be src-dst-ip.

But on the other hand, as written above - the VDS health checks does not complain.

What do you think?

Should I configure the load-balance setting to src-dest-ip ? Will there be a distruption of service?

Thanks!!

[weinstein5]

Since all seems to be working I would not change anything -

[DavidPasek]

It looks LACP configuration from vSphere VDS is correct.

The only "problem" you have is load balancing algorithm on the CISCO side. Please be aware it is not technical problem because if port-channel is established than load balancing algorithm on different sides of the channel can be different and it works. What algorithm you want on the CISCO side must be selected by network admin. To define the load-balance method, enter in the global configuration mode and use the command “port-channel load-balance ?” to see the load-distribution method available.

[VirtuallyMikeB]

Hello,

On the 3750, src-dst-mac hash is the default.  There is no problem if you have different hashing algorithms in the path.  For consistency or standardized configurations, you can force everything to use one hashing algorithm or another, but there's nothing inherently wrong with using multiple types in a path.  Every device in the path always uses its own algorithm to choose which uplink or port to use.  For more static configurations, like IP storage, where source and destination IP and MAC addresses don't change often, your hashes and therefore the uplinks used will be fairly static and unchanging.  If you're using hashing for something like virtual machine traffic, especially where client devices access the VMs, like a file or print server, you'll get more distributed hashes because you have more source-destination pairs.

All the best,

Mike

-----------------------------------------

Please consider marking this answer "correct" or "helpful" if you found it useful.

Mike Brown

VMware, Cisco Data Center, and NetApp dude

Consulting Engineer

[e-mail]

Twitter: @VirtuallyMikeB

Blog: http://VirtuallyMikeBrown.com

LinkedIn: http://LinkedIn.com/in/michaelbbrown