
I'm curious if the slowness you saw vs Parallels was simply on the interface responsiveness side or if it also extended to lowered vCPU performance as well.
Puppy linux running mac os in virtualbox full#
This means never running a full desktop environment unless you're actually building for or testing in that specific DE. If you need to use a graphical application, give it its own VM with as little overhead as possible. If at all possible, run your VMs without X/Wayland. Making sure your VMs are optimized for specific tasks is a big part of getting better performance out of VBox, especially from an interface responsiveness standpoint. KVM is what we use on servers as well, but our developers build for those environments by using VBox VMs on their local workstations (a mix of Mac and Windows). Make sure you have the guest extensions installed. Make sure you're not using graphics acceleration on the desktop (or a desktop environment in general) unless you actually need it. To help with OP's query: Make sure your host hardware supports VT extensions on the CPU.

We've migrated to BSD variants for all of our bare-metal use cases (mostly network stack and host hypervisors) and almost exclusively containerize Linux at this point, with no significant performance degradation. I come down on the opposite side of the coin: there are very few reasons anymore to run Linux on bare metal because it's so heavily optimized for VM/containerized usage at this point.

We have several dozen Debian and RHEL guests running in production on top of VBox, what configurations are you running where performance is a problem? It's not as easy to optimize for as VMWare, but since we have optimized, performance has been functionally identical for us. We're using it extensively and performance for Linux guests is very close to native if the VM and the host are properly configured.
