documentation VS firecracker-containerd

Both Kata Containers and UTM support virtio-fs, so this is not strictly true. The former can be used as a stand-in replacement for the runtime used by docker desktop[1]. With the latter, one could use a UTM-backed guest as a docker runtime in macOS[2] or run docker directly on the guest[3].

[1] https://github.com/kata-containers/documentation/blob/master...

[2] https://www.codeluge.com/post/setting-up-docker-on-macos-m1-...

[3] https://www.lifeintech.com/2021/11/03/docker-performance-on-...

For services with increased security requirements, it is recommended to use a low-level run-time with a high degree of isolation (gVisior, Kata-runtime)

On the official Kata repo, I found a tutorial only for manually deployed Kubernetes on GCE.

This is new and may not be used much, but it is possible to use part of Kata with part of Firecracker. https://github.com/kata-containers/documentation/wiki/Initia...

If it's using firecracker, it's probably using KVM virtualization while ensuring that the memory the VM consumes is not pinned... that is, that the VM can be swapped out of memory. For reference, firecracker was created by AWS to run and secure AWS Lambda. The hypervisor is written in rust and uses seccomp to eliminate unnecessary system calls. They open sourced it a few years back.

What you gain is a stronger security boundary. Just FYI, since 2019, you can also do this in Kubernetes using Kata containers which will happily shim firecracker. The setup is not simple though.

https://github.com/kata-containers/documentation/wiki/Initia...

Overall, fly.io building infrastructure on this pattern is fantastic and making it accessible is fantastic. Looking forward to seeing how this continues to evolve and am happy to see more infra build on top of firecracker. Very exciting!

My team is working on multi-cloud AWS Bottlerocket remix (Azure, GCP) with opt-in support for [firecracker-containerd](https://github.com/firecracker-microvm/firecracker-containerd) for our in-house CNCF distro, investigating microkernels applicability (tldr; they are not production-ready). We test kubernetes compat and migration plans for over 40+ cherry-picked solutions, and facing numerous compat issues for every k8s update. We do have support for Container Managed Control Planes described above, as well.

You could use a container sandbox like gVisor, light virtual machines as containers (Kata containers, firecracker + containerd) or full virtual machines (virtlet as a CRI).

There is this project, which I have never used, but seems promising. https://github.com/firecracker-microvm/firecracker-container...

, "should represent the path to a file that contains a JSON which stores the entire configuration for all of the microVM's resources" (okay this is fair enough). Also, they stipulate, "**The JSON must contain the configuration for the guest kernel and rootfs, as these are mandatory, but all of the other resources are optional, so it's your choice if you want to configure them or not. Because using this configuration method will also start the microVM, you need to specify all desired pre-boot configurable resources in that JSON.**" **File Names for the Pre-Boot Resources** (included within the greater repo here): 1. **firecracker.yaml** - Names of resources are contained here ; 'file nad the names of their fields are the same that are used in API requests' (cool) 2. **tests/framework/vm_config.json** (boilerplate config file to guide us - great) > *"After the machine is booted, you can still use the socket to send API requests for post-boot operations."* (this honestly feels clunky as a mf) ### Conclusion Somewhat of a pain in the ass (just looking through the directions); the fact that we'd have to go grab a uncompressed kernel image + file system image (ext4) is kind of a fucking hassle / burden. Was hoping for a solution more akin to Docker where it can just be spun up real quick & then deployed. But they claim that this 'jailer' feature (that they keep hyping) will **ensure** (I guess?) that whatever is done within the container will remain within the container (and not escape). I haven't seen anything that sticks out about this project that leads me to believe that it possesses that capability, but I definitely don't want to rule it out. ### Extra Documentation + Information 1. **OSv Running on 'Firecracker'** (yay more work though) - http://blog.osv.io/blog/2019/04/19/making-OSv-run-on-firecraker/ 2. **Building OSv Images Using Docker** - http://blog.osv.io/blog/2015/04/27/docker/ 3. **firecracker containerd** (this is something that's probably important for the overall mission of what we want to accomplish here) - https://github.com/firecracker-microvm/firecracker-containerd ### Firecracker Containerd **Description** - "*firecracker-containerd enables containerd to manage containers as Firecracker microVMs*" - "This repository enables the use of a container runtime, containerd, to manage Firecracker microVMs. Like traditional containers, Firecracker microVMs offer fast start-up and shut-down and minimal overhead. Unlike traditional containers, however, they can provide an additional layer of isolation via the KVM hypervisor." **They Also Identify Potential Use-Cases in the Repo Such as** 1. "*Sandbox a partially or fully untrusted third party container in its own microVM. This would reduce the likelihood of leaking secrets via the third party container, for example.*" 2. "*Bin-pack disparate container workloads on the same host, while maintaining a high level of isolation between containers. Because the overhead of Firecracker is low, the achievable container density per host should be comparable to running containers using kernel-based container runtimes, without the isolation compromise of such solutions. Multi-tenant hosts would particularly benefit from this use case.*" Really interesting feature of this repo here is: "*A root file filesystem image builder that constructs a firecracker microVM root filesystem containing runc and the firecracker-containerd agent.*" (that could save a lot of time on that whole filesystem image thing that they were mentioning prior) **Additional Links of Importance** 1. **Getting Started Guide** - https://github.com/firecracker-microvm/firecracker-containerd/blob/main/docs/getting-started.md 2. **Quickstart Guide** - https://github.com/firecracker-microvm/firecracker-containerd/blob/main/docs/quickstart.md 3. **A Root Filesystem Image Builder** - https://github.com/firecracker-microvm/firecracker-containerd/blob/main/tools/image-builder 4. **Runtime Linking Containerd** - https://github.com/firecracker-microvm/firecracker-containerd/blob/main/runtime **Documentation All Located Here** - https://github.com/firecracker-microvm/firecracker-containerd/tree/main/docs (definitely fucking needed because there's a lot here to wrap one's head around) - **Design Approaches Doc** - https://github.com/firecracker-microvm/firecracker-containerd/blob/main/docs/design-approaches.md - **Shim Architecture** - https://github.com/firecracker-microvm/firecracker-containerd/blob/main/docs/shim-design.md - **Launching 4k VMs Using Firecracker** - https://github.com/firecracker-microvm/firecracker-demo - **firectl** (CLI options for manipulating this tool from terminal ; this is important as well) - https://github.com/firecracker-microvm/firectl [damn, there's a lot that came with this here!]

Exactly, it is about secure multi-tennancy. If I recall correctly firecracker doesn't replace containerd, microVMs still runs some sort of it. Anyway, you still need a base OS because container doesn't have the whole OS image. Also I think you can have multiple containers in a single Fargate task so they have to be isolated too.

How does that compare to firecracker-containerd?

https://github.com/firecracker-microvm/firecracker-container...

This repository enables the use of a container runtime, containerd, to manage Firecracker microVMs. Like traditional containers, Firecracker microVMs offer fast start-up and shut-down and minimal overhead. Unlike traditional containers, however, they can provide an additional layer of isolation via the KVM hypervisor.

I'm really impressed by fly.io, and the candidness with which they share some of their really awesome technology. Being container-first is the next step for PaaS IMO and they are ahead of the pack.

I aim to build a platform like theirs someday (probably not any time soon) but I don't think I'd do any of what they're doing -- it feels unnecessary. Bear with me as I recently learned that they use nomad[0] and some of these suggestions are kubernetes projects but I'd love to hear why the following technologies were decided against (if they were):

- kata-containers[1] (it does the whole container -> VM flow for you, automatically, nemu, firecracker) with multiple VMM options[2]

- linuxkit[3] (let's say you didn't go with kata-containers, this is another container->VM path)

- firecracker-containerd[4] (very minimal keep-your-container-but-run-it-as-a-VM)

- kubevirt[5] (if you just want to actually run VMs, regardless of how you built them)

- Ceph[6] for storage -- make LVM pools and just give them to Ceph, you'll get blocks, distributed filesystems (CephFS), and object gateways (S3/Swift) out of it (in the k8s space Rook manages this)

As an aside to all this, there's also LXD, which supports running "system" (user namespace isolated) containers, VMs (somewhat recent[7][8]), live migration via criu[9], management/migration of underlying filesystems, runs on LVM or zfs[10], it's basically all-in-one, but does fall behind in terms of ecosystem since everyone else is aboard the "cloud native"/"works-with-kubernetes" train.

I've basically how I plan to run a service like fly.io if I ever did -- so maybe my secret is out, but I sure would like to know just how much of this fly.io got built on (if any of it), and/or what was turned down.

[0]: https://news.ycombinator.com/item?id=26745514

[1]: https://github.com/kata-containers/kata-containers

[2]: https://github.com/kata-containers/kata-containers/blob/2fc7...

[3]: https://github.com/linuxkit/linuxkit

[4]: https://github.com/firecracker-microvm/firecracker-container...

[5]: https://github.com/kubevirt/kubevirt

[6]: https://docs.ceph.com/

[7]: https://discuss.linuxcontainers.org/t/running-virtual-machin...

[8]: https://github.com/lxc/lxd/issues/6205

[9]: https://criu.org/Main_Page

[10]: https://linuxcontainers.org/lxd/docs/master/storage

https://github.com/firecracker-microvm/firecracker-container...