Using LACP with a vSphere Distributed Switch 5.1

by Chris Wahl on Oct 15th, 2012 | 6,347 views

One of the more exciting and eagerly anticipated announcements in vSphere 5.1 (at least for me) were all the distributed switch enhancements. Among the new feature list is the ability to use LACP (Link Aggregation Control Protocol) on a distributed switch. Formerly, the switch was limited to a static EtherChannel or use of a third party switch, such as the Cisco 1000v, if LACP was desired. This post will go into how to actually configure and enable LACP on the latest vSphere Distributed Switch 5.1.

Want more details on the new features? Check out my Deep Dive on vSphere Distributed Switch 5.1 Features posts.

vSphere Distributed Switch Configuration

To start with, you’ll need a VDS running version 5.1. Additionally, you’ll need to pay heed to the list of caveats below:

vSphere supports only one LACP group per distributed switch and only one LACP group per host.
LACP does not support Port mirroring
LACP settings do not exist in host profiles
LACP only works with IP Hash load balancing and Link Status Network failover detection
LACP between two nested ESXi hosts is not possible

Source KB

I highlighted the last one for those who are running nested ESXi labs. Sorry!

In my lab environment, I have created a VDS named “VDS-LACP” and a single port group named “LACP-TEST”. Make sure to set the load balancing method to IP Hash and ensure that all uplinks are active.

The LACP-TEST port group configuration details

One additional step remains: you must enable LACP on the uplink port group. This must be done via the vSphere Web Client. Navigate to the uplink port group, click the Manage tab, and then select the Settings button and click Edit.

Have no fear, I’ve drawn you a GUI map!

Uplinks > Manage > Settings > Properties > Edit … say that 3 times fast

Once there, enable LACP and set it to Active.

Physical Switch Configuration

Ensure that your vSphere uplinks are plumbed into a physical switch that is configured properly for LACP. In my case, this means creating a Link Aggregation Group (LAG) on my HP V1910 switch in the lab, shown below:

The LAG creation screen on my V1910

Note that I have clicked the radio button for “Dynamic (LACP Enabled)” for the interface type. For instructions on how to create a static EtherChannel, you can refer to this video “Creating a Link Aggregation Group for a vSphere Lab” that I posted a while back.

Results

If you’ve done everything right, the LACP status should show all members as active. If you forgot to enable LACP on the VDS uplinks, you’ll get an error and one or more of the ports will be in standby mode as shown below:

This most likely means that you did not enable LACP on the VDS uplink port group

You’ll also get a neat new item on the ESXTOP list called “LACP_MgmtPort”. I assume this handles the LACP negotiation on behalf of the VDS.

Thoughts

I’m really glad to see LACP join the ranks of native features in the VDS, and am a little surprised it took this long to get it included. After all, this is a feature limited to those with the highest licensing level: Enterprise Plus. I haven’t had a lot of stick time with the feature yet so I can’t comment on any pros or cons from a troubleshooting or reliability standpoint, but the configuration process was simple enough.

If you are running, or plan to run, LACP on your VDS 5.1 and care to share your experiences please leave a comment below!