Multi-cloud RAID. Not multi-cloud sync.

Why multi-cloud RAID beats multi-cloud sync

The dominant "multi-cloud storage" approach replicates your file to each provider — each copy is independently readable. A breach of any one provider exposes everything. The "multi-cloud" of replication buys availability; it does not buy security against single-provider compromise.

Erasure coding is a different mathematical object. RS(5,2) splits each frame into seven shards. Any five reconstruct; fewer than five reveal nothing about the original. No single provider — and no two providers — can read your file. The provider you trust least is no more dangerous than the provider you trust most.

Storage economics

RS(5,2) carries 7/5 = 1.4× storage overhead. Triple-replication carries 3× overhead. At petabyte scale, the difference compounds: RS(5,2) buys the same failure tolerance as triple-replication at less than half the storage cost — plus the security property that replication does not give you at any cost.

Failure isolation

Reads return as soon as any five of seven shards succeed. Two slow providers don't affect read latency. Two failed providers don't affect read availability. Three failed providers force a degraded read with explicit error semantics. This is meaningfully more resilient than the typical replicated-multi-cloud setup, which degrades to single-cloud on the second failure.

Bring your own providers

Fourteen plugins ship out of the box. The plugin SDK lets you integrate a sovereign or regional provider in under 500 lines of code. The CloudProvider interface is the same across all plugins: upload, download, delete, health-check, region-detect. Want to keep some shards on your existing on-prem MinIO and some on a sovereign French provider not in our catalogue? Integrate both as plugins; the RAID layer is provider-agnostic.

Continuous connexion health

Provider heartbeats run roughly every five minutes. Staleness, latency, quota, and credential health surface before they affect uploads. Stale OAuth credentials get a badge on the connexion the moment a token drifts. An aggregated connexion-issues banner appears on the file browser, so operators learn about provider problems before failed writes rather than from them.

Dedup analysis

The platform detects identical-content files across the tenant — cross-account where scoped — surfaces a savings report, and lets administrators reclaim storage by hard-linking the redundancy. For organisations with significant duplication (typical: 15-30% of total volume) the savings recoup meaningful portions of storage cost without changing user behaviour.