Prove it before it ships

Every release must pass tests where real clusters are made to fail.

What we offer

Test and CI build-out

Before your distributed system goes to production, we build the fault-injection tests, CI pipelines and release checks it needs.

6–12 weeks

Performance regression

Benchmarks re-run on every change, including while recovery is in progress.

DevOps & test automation

Proven in the lab, ready for production

Distributed systems fail in ways that only real clusters reveal. We build deterministic automation that stands up real multi-node clusters, injects exact failures, and gates every release on the result.

Release pipeline for a distributed storage system

  1. Provision

    Multi-node clusters built from code: VMs, containers or a physical lab.

  2. Build matrix

    Kernel-matched builds across distros and architectures, signed packages.

  3. Functional

    POSIX, protocol and API suites across every supported configuration.

  4. Fault injection

    We kill nodes, disks and links on purpose while I/O is running.

  5. Perf regression

    Benchmarks re-run on every change, including while recovery is in progress.

  6. Release gate

    Integrity and performance results decide go / no-go.

Failure scenarios encoded in tests
ScenarioWhat we verify
Node crashClients reconnect, no acknowledged write is lost
Network partitionConsensus holds, no split-brain, clean rejoin
Disk failureChunk recovery completes, every checksum matches
Failure mid-resyncMirrored metadata survives a second failure
Live upgradeTwo major versions mount side by side; rollback is safe
Pod restartCSI block devices keep serving in-flight I/O

Deterministic test frameworks

Each test records exact fault settings, so every failure can be reproduced. The test frameworks have their own tests too.

  • Ansible
  • pytest
  • Python

Formal verification

Model-check consensus and protocol logic before it is written in C.

  • TLA+
  • TLC

CI on real hardware

Native ARM and x86 runners, Vagrant VMs and a physical lab, as well as containers.

  • CI runners
  • Vagrant
  • Docker

Storage slower than your compute?

Send us your topology, workload and a benchmark run. We will get back to you.