How to size Azure Kubernetes for Elastic deployment

We are currently running a 50 node cluster (with 40 data nodes) on VMs in Azure. I am looking into converting to running in AKS. I am having trouble figuring out how to determine the resources I need in my AKS cluster to host the ES nodes. My cluster is current running on Standard_E8s_v3 machines (8 cpu / 64 gb). I am not clear on how to calculate how many AKS nodes I need to be able to run a 50 ES nodes. Is it a 1:1 mapping (i.e., I would need 50 AKS nodes), or is running in containers more efficient and each ES node won't need full utilization of all 8 cores, etc.?


Unfortunately I think it is hard to give an accurate answer. It depends on the underlying implementations of the hypervisor and the container runtime. Having a 1:1 mapping seems to be a good start, but I guess that you'll have to do some tests to understand if there are significant performance discrepancies between VMs and containers.

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.