Week 1 — landing scaffold. Live playground: weeks 3–4.

Sub-second JVM cold-start
for enterprise Java.

KVM microVM snapshot/restore plus MAP_PRIVATE fork-many. Boot Spring Boot once, fork thousands of ready-to-serve replicas in ~200 ms each. No code changes. Any JDK. Your database wherever it lives.

Fork latency, p50
~200 ms
Spring Boot single-jar, warmed snapshot
Per-fork memory
~0.5 MiB
Pss, thanks to MAP_PRIVATE CoW
Cold-boot baseline
8–30 s
What snapshot/fork replaces

Progress

Stars
5
Forks
1
Open issues
5
Last commit
41 min ago

Try it live

Each visitor will click a button and get their own real Spring Petclinic instance, running inside a nanovm-forked KVM guest, reachable at a personal URL for ~10 minutes. Real HTTP, real Eureka lookup, real database queries. Not a screencast, not a recording.

  1. Completed: Week 1Landing page + Fly.io deployment pipeline
  2. Upcoming: Week 2virtio-net + host bridge (real network into each guest)
  3. Upcoming: Week 3First live spawn — one Petclinic per visitor, cold boot
  4. Upcoming: Week 4Warm pool + snapshot/fork. Sub-second spawn
  5. Upcoming: Week 5Microservices variant: full 7-service stack per visitor
  6. Upcoming: Week 6Live metrics dashboard + enterprise trial contact form