VDS Load Balancing for LAG / LACP

[MJMSRI]

Hi all, I want to check the correct load balancing mode for the setup below.

• 1 x Distributed Switch with 2 x 25gb uplinks
• uplinks connected into 2 Cisco catalyst switches
• LACP setup on Cisco catalyst switches for both interfaces on each host in Active mode.
• distributed switch setup with LAG and both 25gb uplinks added into LAG obn VDS
• LAG set as Active uplink on VDS Port Groups. The 2 uplinks set as unused.

I have found the load balancing set current as “use explicit failover order”.

I believe the load balancing above is wrong and should be set as “Route based on IP Hash”, is this correct?

Looking at this article: https://kb.vmware.com/s/article/1004048 it simply details: “From the Load Balancing dropdown, select the correct load balancing policy. This will be determined by the physical switch. Refer to the physical switch vendor if there are questions on which load balancing algorithm should be used.”

 

[Maarten Van Driessen]

You are correct, when using LACP, you need to set the load balancing algorithm to IP Hash.

That being said, I find that LACP adds very little besides complexity and I'm a strong advocate for not using LACP on your ESXi hosts.

[endriopettini]

It depends heavily on your environment; if your traffic is primarily at the L2 level, it makes more sense to use the MAC-based algorithm as it involves one less level of processing. Some devices offer the option to use both IP and MAC, which is another way to effectively balance traffic for both layers of the OSI model (L2 and L3).