Resistro Cloud provides a backup baseline, monitoring, and restore evidence for small self-managed Postgres setups without an internal backup-ops team.
Postgres is the product anchor. MySQL/MariaDB are bounded logical-dump paths, not PITR or physical-backup parity.
Resistro started as an open-source CLI and engine, not as a SaaS dashboard. Resistro Cloud adds the operating layer for small Postgres setups: setup review, agent attachment, monitoring, restore drills, and evidence. MySQL/MariaDB remain request-only boundary cases without parity claims.
The community platform behind our amateur-radio project, 11m-band.de, stores its live PostgreSQL database as an ordinary Resistro Cloud tenant — the same product path we sell, not a separate internal setup. Backups run on an hourly cadence, monitored for silent failures, using the same agent, the same encryption, and the same default storage path through Hetzner data centers in Germany/EU every customer gets. If the restore path broke, we would be the first to notice.
A customer-triggered restore drill produces a dated result record and a restore-evidence PDF: selected backup, restore target, outcome, and operator notes. It is evidence for the tested restore path, not a legal certification or SLA.
Software, agent attachment, and monitoring instead of a managed-retainer promise.
Daily backups on Solo and Starter, hourly from Business. The agent asks for work on a 1h default interval on every plan; the plan decides how often a backup is actually taken.
Request-only for self-managed databases; no parity, PITR, or physical backup promise.
PITR is proven, but not a current product promise, not a continuous guarantee, and not a hard RTO/RPO commitment.
Restore tests as ongoing operating checks for the defined restore path.
Customer-triggered drills with restore-evidence PDF and operator-assisted result handling.
AES-256-GCM client-side before upload; the server receives only a SHA-256 hash of the key, never the key or plaintext data.
Hetzner data centers in Germany/EU; backups are stored encrypted. Roles, subprocessors, and transfers are reviewed in the privacy/AVV package.
Alerting after roughly one day; current monitor threshold: 28h.
Daily `/healthz` smoke test for public product paths.
These items are intentionally marked as roadmap. They are not part of today's operating promise, but they show which procurement questions Resistro needs to close next.
Today, backup keys stay on the customer side. The next enterprise step is a clear BYOK flow with auditable rotation, key status, and explicit responsibility when a key is lost.
SOC 2 Type II is the evidence enterprise and regulated buyers expect for continuously operated controls. Resistro treats SOC 2 Type II as roadmap, not as an existing certification.
Resistro Cloud is for small self-managed Postgres setups, typically 1 to 10 databases: solo developers, small agencies, and small SaaS teams on their own Linux host. A daily backup on Solo or Starter — or hourly from Business — is enough, restore evidence matters, and there is no internal backup-ops team.
A Postgres backup baseline with a plan-based cadence — daily on Solo and Starter, hourly from Business — AES-GCM encryption, customer-held keys, default storage through Hetzner data centers in Germany/EU, a daily restore-path check, stale-backup alerting after roughly one day with a current 28h monitor threshold, and a daily `/healthz` smoke test. MySQL/MariaDB is request-only without a parity promise.
No. The daily restore test is an operating check of the defined restore path. It ensures the restore mechanism does not silently break, but it does not mean every customer database is individually validated every day in a full emergency scenario.
PITR is proven for Postgres, but it is not a current product promise. There is no continuous PITR guarantee, no real-size claim, and no hard RTO/RPO. MySQL/MariaDB PITR is not covered.
Storage location is not the whole jurisdiction question. The default storage path uses Hetzner data centers in Germany/EU; backups are stored encrypted. AVV/TOMs, subprocessors, and third-country transfers need to be reviewed in the privacy package before customer use. This is not legal advice and not a promise about authority-access or third-country-transfer risks.
It follows the plan: around one day on Solo and Starter, around one hour from Business. The agent asks for backup work on a 1h default interval on every plan, but the active plan decides how often a backup is accepted. The target is derived from that cadence and is not a hard guarantee.
Restore is manual and operator-assisted. There is no self-service button. Customers can trigger restore drills on demand to verify the restore path and download a restore-evidence PDF.
The restore-evidence PDF documents the time, scope, and result of a concrete restore drill. It is structured evidence for the tested restore path, not a legal certification and not a compliance certificate.
No. There is no support guarantee, uptime guarantee, service-credit model, or fixed RTO commitment.
Resistro uses heartbeat- and Prometheus-based monitoring. The current stale-backup alert threshold is 28 hours without a successful backup signal.
Backup data is encrypted with AES-GCM and the key stays on the customer side. That is the entire promise, including no key recovery if the customer loses the key.
Hosted or managed database platforms such as managed MySQL/MariaDB, RDS, or Supabase, requirements outside the defined small Postgres ICP, and any MySQL/MariaDB requirement beyond bounded logical dumps. MySQL/MariaDB-heavy is not an equal primary fit.
No. It verifies basic reachability of central public product paths, not a full uptime or end-to-end guarantee.
Because Resistro Cloud sells only what can be narrowly evidenced today: bounded promises, restore verification, and explicit limits instead of broad security rhetoric or implied guarantees.
Resistro Cloud fits when you run self-managed Postgres databases, do not have an internal backup-ops team, and need restore verification as an operating process. MySQL/MariaDB fits only as bounded logical-dump support. It does not fit when you need hard uptime/support guarantees, hosted database operations, or fully automated self-service restore.