Clone prod in a minute. Roll back in five. Stop filing tickets.
CLI-first infrastructure that clones production in 60 seconds, rolls back in under five minutes, and runs on self-owned hardware. No tickets. No waiting. The data never leaves.
Everything is automated. Except getting a test environment.
CI/CD deploys in minutes. IaC provisions in seconds. But a test environment with real data means filing a ticket and waiting three days. Meanwhile, staging is shared, flaky, and nothing like production.
See the timeline
One command. Full production clone.
Clone
Run rdc ceph snapshot create production. Btrfs CoW takes a sub-second snapshot of everything: app, database, config.
Work
Start coding immediately against production-identical data. Push a branch, get a fresh clone.
Recover
Something breaks? Roll back in seconds. Immutable snapshots mean no data is ever lost.
Why 60-second clones are possible
Rediacc uses btrfs copy-on-write snapshots at the filesystem level. A snapshot of a 380 GB production environment completes in sub-seconds and consumes zero additional storage initially. As the clone diverges, only changed blocks use new space.
Why 60-second clones are possible
| Traditional approach | Rediacc (btrfs CoW) |
|---|---|
| Docker volume copy: full data copy, minutes to hours | Metadata-only snapshot: sub-second, any size |
| pg_dump 380 GB: 12-36 hours, one DB engine at a time | Any database cloned in one atomic operation |
| 10 clones = 10x storage (3.8 TB) | 10 clones share blocks via CoW (~15 GB delta) |
| Rollback: restore from backup (30+ min) | Rollback: btrfs snapshot switch (<5 min) |
| Cloning a Kubernetes cluster: re-provision, redeploy, re-seed data | rdc cluster fork copies it, data included. 46 seconds in our lab |
Built for engineers who automate everything
No ticket, no queue
Getting an environment stops being someone else's sprint item. It is yours to create and to throw away, as often as the work actually needs one.
It scripts like anything else
Every operation is a command with an exit code, so it drops into a pipeline, a Git hook or a Makefile without a wrapper, a browser or a service account.
Mistakes stop being expensive
A bad migration or a wrong DROP is undone by switching back to the snapshot, which changes how freely people are willing to try things against real data.
Related solutions
Try it
Install the CLI. Clone the first environment. Start the 14-day free trial, no sales call.
Start free trial