03Technology
Storage.
Where the state actually lives: ZFS-backed storage with end-to-end integrity, replication engineered against real failure scenarios, and backups that have been restored, not just taken.
A backup is
a hypothesis.
Until a restore has been exercised, a backup is a hypothesis with good intentions. NetFRAME designs storage from the recovery path backwards: what must survive, how it returns, what that return path depends on, and how long it takes when it matters.
ZFS anchors the stack because it makes integrity a property rather than a hope: checksummed end to end, snapshotted cheaply, and replicated with verifiable results.
01
Integrity & replication
ZFS provides checksummed storage with self-healing redundancy, cheap snapshots, and send/receive replication. Pool topology, redundancy level, and snapshot cadence are engineering decisions made against workload and failure requirements.
Replication targets are placed in separate failure domains, because a replica that shares power, cooling, or a chassis with its source is bookkeeping, not protection.
02
Backup architecture
Backup layered above replication with independent tooling and retention, so a logical corruption replicated faithfully can still be escaped. Restore paths are dependency-checked against the disasters they exist for and rehearsed on a schedule.
Storage telemetry (capacity, latency, scrub and error state) feeds the estate observability stack, so degradation is a trend line before it is an incident.
Technologies we work with
- ZFSChecksummed pools, snapshots, and replication.
- Snapshot strategyCadence and retention engineered per workload.
- Backup toolingIndependent layers with rehearsed restores.
- Storage telemetryCapacity and health as observable trend lines.
Related