Writing
Notes, mostly boring.
Protocols, .NET internals, and infrastructure that misbehaves.
Firecracker, part four: BusyBox init, SquashFS, and a day lost to systemd
Replacing a fragile init, building rootfs images in Docker, and ending up exactly where I started — but better.
Firecracker, part three: a request's journey through MangoFaaS
Gateway to Kafka to a microVM and back over vsock — with the first cold-start numbers.
Firecracker, part two: one rootfs, many microVMs
OverlayFS, sparse writable layers, and how to stop a not-so-micro VM running out of disk.
Firecracker, part one: getting a microVM to boot
What Firecracker actually is, why it isn't a container, and the unglamorous work of a first boot.
Get in touch
[email protected]A CV is available on request — I share it privately rather than publishing it.