Best practices for standing up Atlassian Platform using Amazon Kubernetes Cluster

[D M]

Hi,

I was working on standing up the suite of Atlassian applications: namely Jira, Confluence, Bamboo, Crowd, and Bitbucket.

The goal is to utilize AWS's Elastic Kubernetes Service as the deployment platform for these apps and manage that via Helm Charts. But, my question is in regards to the infrastructure itself and what the best practices surrounding it is.

With the way that Kubernetes (and specifically AWS) is set up, would it make sense to deploy all of the Atlassian applications to the same EKS cluster or should I be consider separating them to separate clusters? (Context: I'm new to setting up Atlassian apps including Data Center)

I know by deploying all of the apps to the same EKS cluster it would save on the overhead of managing the cluster itself, but I wasn't sure if this would cause any limitations on resource utilization or competing for resources amongst the applications themselves. 

General guidance on how to stand up the Atlassian Platform on a large scale into an EKS cluster would be greatly appreciated! Thanks in advance.

[Yevhen]

@D M I personally don't see anything wrong with deploying Atlassian DC products to the same cluster provided that each product has its own nodegroup and you do a good job with setting reasonable resources requests/limits, use affinity/anti-affinity rules, tolerations etc.

Atlassian DC products and DC Helm charts do not make any assumptions on the k8s clusters, clouds and providers.