Idealy memory requirements when Harbor is deployed on K8S Cluster

[prashi2202]

Hi Experts,

We have currently deployed Harbor on K8S cluster and would like to know from you all as to what is the ideal memory and CPU requirements, our target clients are expected to push large images which might go up to 100GB.

Would like to know what is the ideal memory and CPU configuration for such requirement.

Also need help on understanding how Jobservice, Core and Registry pods are interconnected with regards to memory management.

Current configuration is as below.

<style> </style>

Pods	PROD - Memory Limits in MB	CPU
nginx	128	100m
portal	128	100m
core	512	100m
jobservice	512	100m
registry	3072	100m
registry-controller	3072	100m
chartmuseum	128	100m
notary	128	100m
redis	128	100m
exporter	128	100m
trivy	1024	200m
Total	8960	1200m

[heww]

1200m means 1.2 core CPU, it's small for the normal usage of the harbor.

The resource used by the harbor depends on the requests rate and network bandwidth, you can measure the resources using https://github.com/goharbor/perf

The registry uses a lot of memory when pulling/pushing large images.