I have a customer who recently told me they are ripping out the LACP configuration on their vSphere hosts. I asked them why and they simply stated…
“It’s difficult to configure and deploy, additionally we want to start leveraging vSphere Auto-Deploy and host profiles for deploying new hosts!“
This prompted me to set off on the journey of configuring LACP in my lab and see what’s to this “hard to configure stuff”. Well it turns out that it can be a bit of a challenge, especially if you cannot have any downtime during the migration. I attempted several methods and failed a ton of times but ultimately got it. Frankly it wasn’t that difficult to do, but it was very time consuming to do it and ensure absolutely no down time.
There are a ton of benefits to LACP for sure, bundling together of multiple physical links, more efficient network redundancy and failover, and increased available bandwidth. These are big time pros to using Link aggregation (LAG) in accordance with IEEE 802.1AX-2008 (previously IEEE 802.3ad).
Within vSphere there are also a ton of limitations, especially not being able to use host profiles to configure this stuff quickly when adding new hosts.
In summary, should you deploy Enhanced LACP in your environment? Well that all depends doesn’t it, I for one will leave it configured in my lab (love the benefits), but I do not have the task of configuring 100s or 1000s of ESXi hosts, or working with different teams to get this deployed. Can you imagine what it takes to get this deployed in an Enterprise production environment with a facilities team, a networking team, and the virtual infrastructure team?
You might want to just say no to LACP and stick with good old Route Based on Originating Virtual Port ID or a personal favorite Route Based on Physical NIC Load. Good luck!