Hello everyone,
I’m looking for insights on the limitations of using Horizontal Pod Autoscaler (HPA) with databases running as StatefulSets. Specifically, I want to understand the challenges and best practices related to scaling stateful applications like databases in Kubernetes.
If anyone has experience or can point me to relevant documentation, forum discussions, or resources, I would greatly appreciate it!
Thanks in advance!
As you get deep into the world of StatefulSets its gets more complex and it needs right orchestration and dependency mechanism to function in efficient manner.
Some of the challenges with it is losing the data specially when the delete/modify process might take too much time or abrupts intermittently due to network issues.
This is the reason why many organization prefer vendor based operator to manage it instead of rewriting it. e.g. Kafka has its own operator, same is for mysql
In my opinion it’s core concept is more about that specific software (e.g. mysql) works so building or using existing more widely used CRD/operator is more better option issue to resolve common issue associated with it as open source community.
relevant documentation, forum discussions, or resources
That would depend on what database you planning to use, I believe each one will have its own community for k8s operator specific discussions.