In production we're going to deploy a redis server and need to set the overcommit_memory and disable Transparent Huge Pages in Kernel.
The issue is currently we only have one giant server, and it is to be shared by many other apps. We only want those kernel configs in the redis server. I wonder if we can achieve it by spinning up a dedicated VM for redis. Doing so in docker certainly doesn't make sense. My questions are:
- Will those Kernel configs take actual effect in the redis VM even if the host OS doesn't have the same configs? I doubt it since the hardware resource is allocated by the host machine in the end.
- Will the kernel config in the redis VM affect other VMs that run other apps? I think it won't, just want to confirm.
- To achieve the goal, what kind of VM or hypervisor should we use?
If there's no way to do it in VM, is having a separate server (hardware) for redis the only way to go?