[Best Practice] when creating multiple nodes (for clustered instances)

Knowledge Drop

Last tested: Jan 12, 2021
 

Scenario:

You are looking to expand the number of nodes hosting Looker from 2 to 5 and the two current nodes have 96GB memory and you are adding three more nodes with 24GB of memory.

Question:

Does this make a difference overall? Is it best practice to have the same memory allocation distributed across all nodes?

Answer:

It is best practice to keep all the instances (nodes) identical in a cluster. This is because Looker won't take into account the memory/cpu of each node, and will treat them all evenly. So if the nodes are different sizes, a smaller node won't be taking a proportionally smaller load from the looker application - The load will be shared equally across nodes, regardless of size. We have seen issues regarding uneven node sizes when scheduling work.

This content is subject to limited support.                

Version history
Last update:
‎07-07-2021 12:36 PM
Updated by: