I’m slightly sorry it’s not OpenBSD’s vmm, but glad there is an option here!
Considering how many problems can be solved by running a different OS in a hypervisor, this means Linux, OSX, Windows, Open, Net, Free, Dragonfly, and apparently soon Haiku, are all possible options for a developer’s personal or work machine.
This is good. It won’t kill the monoculture towards Linux, but it allows those who want to live a little differently the ability to do so, and with more developers than ever, we sure have the developers to send off beaten paths.
Does anyone have a good comparison of nvmm (NetBSD), vmm (OpenBSD) and bhyve (FreeBSD)? I believe NVMM (like Linux’ KVM) requries qemu and provides just a very thin shim layer, whereas bhyve provides its own minimal implementation of core devices. Bhyve has been ported to macOS as xhyve, where it is used by Docker to run Linux VMs, but sits on top of Apple’s Hypervisor.framework instead of the FreeBSD kernel module. I’d love to read an in-depth comparison of the different design philosophies and functionality.
Mind you, I’d also love to have a minimal Type-1 hypervisor. Even the smallest BSDs have a huge kernel that it would be great to get out of guests’ TCBs. Xen is very far from minimal.
I actually kinda wish that FreeBSD had gone the route of a qemu backend shim, similar to NetBSD’s nvmm. That would enable wider adoption, and more features. No need to reimplement the different virtio drivers.
That said, I’m also grateful for bhyve due to how lightweight it is. I’m grateful that it’s BSD licensed as well, enabling inclusion in base. The qemu shim itself could reside in base, but not qemu itself.
I’m slightly sorry it’s not OpenBSD’s vmm, but glad there is an option here!
Considering how many problems can be solved by running a different OS in a hypervisor, this means Linux, OSX, Windows, Open, Net, Free, Dragonfly, and apparently soon Haiku, are all possible options for a developer’s personal or work machine.
This is good. It won’t kill the monoculture towards Linux, but it allows those who want to live a little differently the ability to do so, and with more developers than ever, we sure have the developers to send off beaten paths.
Does anyone have a good comparison of nvmm (NetBSD), vmm (OpenBSD) and bhyve (FreeBSD)? I believe NVMM (like Linux’ KVM) requries qemu and provides just a very thin shim layer, whereas bhyve provides its own minimal implementation of core devices. Bhyve has been ported to macOS as xhyve, where it is used by Docker to run Linux VMs, but sits on top of Apple’s Hypervisor.framework instead of the FreeBSD kernel module. I’d love to read an in-depth comparison of the different design philosophies and functionality.
Mind you, I’d also love to have a minimal Type-1 hypervisor. Even the smallest BSDs have a huge kernel that it would be great to get out of guests’ TCBs. Xen is very far from minimal.
I actually kinda wish that FreeBSD had gone the route of a qemu backend shim, similar to NetBSD’s nvmm. That would enable wider adoption, and more features. No need to reimplement the different virtio drivers.
That said, I’m also grateful for bhyve due to how lightweight it is. I’m grateful that it’s BSD licensed as well, enabling inclusion in base. The qemu shim itself could reside in base, but not qemu itself.