Hello, I want to run some batch, bounded SQL queries from users on-demand, where most of them takes under a second to run, so I want to optimize the time it starts to create a mini local cluster. I use a local execution environment already. What are the things I should keep in mind? I know that's not the intended use case but leveraging Flink like this would benefit us a lot.

It’s pretty common to use a session cluster for workloads like this.

Thanks for the pointer, will look into that. Wouldn't creating a local cluster would be faster? The reason I ask is queries take 100-500ms to run, and the startup overhead becomes significant.

can someone tell me more about it ? how to set it up on k8s ?