Currently, the resource requirements for the RHOSAK components is high when it is installed in the hybrid cloud. Plus, in the future, there is a need to install minimal components that satisfy RHOC installation on a separate cluster but do not consume any RHOSAK components where the customer knows there are not going to be any Kafka instances.

Need to investigate

• How to reduce the number of operator replicas, which will be worked under MGDSTRM-11134, but won't use HPA MGDSTRM-11100
• How to reduce the strimzi resources, which is primarily discussed under MGDSTRM-9976 - has three possible outcomes:
  • Simply reduce the cpu request 
  • Use VPA MGDSTRM-11132
  • Use FSO based logic to manipulate the resources, in the same way as MGDSTRM-11134
• Possibility to not instantiate any ingress controllers until the very first Kafka instance is deployed: MGDSTRM-11101
• Or Remove ingress controller to favor load balancers as cost semantics may have changed
  • The last time this was looked into the conclusion was to eventually use the gateway api: https://docs.google.com/document/d/1gnsgfZVD_31ZG1eCYXpf0nSwYhBqVGQDebqjTQHetQs/edit#  That recommendation will likely still stand as long as we need to account for developer / trial with a single mechanism.