FreeBSD comes to Amazon's lightweight hypervisor

The FreeBSD developer who brought FreeBSD to Amazon EC² has now sewage it moving connected nan company's lightweight Firecracker hypervisor.

"AWS Community Hero" Colin Percival is nan developer of nan Tarsnap online backup work and the portsnap instrumentality for updating FreeBSD. It's acknowledgment to his efforts that FreeBSD is supported connected Amazon EC².

Now, he has announced that FreeBSD supports different caller platform: Amazon's Firecracker hypervisor. The Reg covered Firecracker erstwhile it was announced, and erstwhile Amazon cut nan value of nan Fargate "serverless" platform that Firecracker supports. It's derived from nan lightweight Rust-based crosvm which is part of Google ChromeOS.

This has meant making aggregate adjustments to nan FreeBSD kernel to let for restricted group of services that Firecracker offers to its alleged "microVMs." For instance, Firecracker doesn't support ACPI and makes extended usage of Virtio. While FreeBSD already supported Virtio, it did truthful via ACPI calls, truthful that needed a rewrite.

Firecracker besides informs nan impermanent OS astir nan fistful of devices successful nonstandard ways: it was only really built to support Linux guests, though nan OSv unikernel has besides been ported to it.

The caller FreeBSD support besides meant changes to nan hypervisor itself, specified arsenic incorporating existing patches to amended Firecracker's support of Xen-style PVH booting.

When we described Firecracker, we drew comparisons to different lightweight hypervisors specified arsenic Kata containers – which aren't really containers – and Google's gVisor, a kernel designed to tally wrong a instrumentality to make it much for illustration a VM.

• NetBSD 9.3: A 2022 OS that tin tally connected late-1980s hardware
• Disentangling nan Debian derivatives: Which should you use?
• FreeBSD 13.1 is retired for everything from PowerPC to x86-64
• The sad authorities of Linux desktop diversity: 21 environments, conscionable 2 designs

Eleven years ago, erstwhile nan Reg offered a little history of virtualisation (and its second, third, fourth, and fifth installments), location was a clear favoritism betwixt hypervisors and nan then-still-new-to-Linux tech of containers. A hypervisor emulates a full computer, and runs an full OS successful each virtual machine, while containers each stock nan aforesaid kernel and tally abstracted userlands connected top.

Now, lightweight hypervisors and microVMs are blurring nan lines betwixt nan two, arsenic we described precocious erstwhile talking astir encrypting Kubernetes clusters. A microVM is an OS that knows it's moving arsenic a impermanent wrong different OS, and which has been designed to talk to a hypervisor's services truthful that it doesn't request emulated hardware. The main ideas are that nan impermanent tin beryllium overmuch smaller, and tin commencement overmuch faster.

The flipside of this is "containervisors", notably Ubuntu's LXD. From comments successful various forums, this seems to beryllium 1 of Canonical's much celebrated and well-liked technologies. As opposed to Docker and its ilk, which attraction connected moving a azygous app isolated successful each container, LXD is explicitly designed to tally an full Linux OS (excluding nan kernel) wrong its containers, complete pinch its ain init system. LXD moreover supports dedicating hardware to each container.

In different words, while containers are getting bigger and much complicated, VMs are shrinking down to rival nan size and motorboat velocity of containers. The much different OSes this supports, nan better, we feel.