container API Gateway resources requirements

Products

• CA API Gateway

Issue/Introduction

We are starting our container API gateway configuration in openshift 4.

Currently we are using CentOS OVAs. Each one has 8 cores and 64 RAM GB. Java heap is about 16GB. We have 8 OVAs for our production cluster.

Each node is serving in prod about 10MM execs/day with 1000 execs/s as peaks.

How could we calculate the number of milicores and RAM to request openshift platform for each container?

Environment

Release : 10.1

Resolution

We recommend that you go with the default resource allocation in the production-values.yml file for our gateway Helm chart.

The maximum resource allocation which is limited by licensing (e.g. 6 cores and 8GB memory per container gateway instance with our bundled SKU) until you find the optimal resource allocation for your use cases in your environment (which can vary based on many factors; you can find that your optimal resource allocation is less than the licensed limit).

Below are some Openshift links which you can follow for best practices while determining the resource allocation.

Node resources configuring - https://docs.okd.io/latest/nodes/nodes/nodes-nodes-resources-configuring.html
Overcommitting - https://docs.okd.io/latest/nodes/clusters/nodes-cluster-overcommit.html
Quota and Limit Ranges - https://docs.okd.io/latest/nodes/clusters/nodes-cluster-limit-ranges.html