Skip to main content

Evaluating SnoutData Cloud

The questions worth asking before you put production data on a young service, answered here rather than left for you to dig out. Each answer links to the page with the detail. SnoutData Cloud opened on 2026-09-04.

Where does my data live?​

AWS us-west-2 (Oregon), on machines we run ourselves. Backups are replicated continuously to us-east-1. There is no EU region, and no other region today. See security and limits.

How much can I lose, and how fast does it come back?​

Recovery point (RPO)With two-zone writes, zero while the copy in another zone is up (driven with the disk destroyed under load: no confirmed write lost). On one zone, every write is on the server's disk at once and in object storage within a minute, and a failed server keeps its disk; only losing the disk or its zone could cost up to about a minute
Two-zone writesPro and Business: a copy in another zone before a commit is confirmed; if that copy is down, writes wait up to 20 seconds, then go on as on one zone until it is back
Recovery time (RTO)1.1 to 1.4 seconds with the data on the machine; about 10 to 21 seconds from object storage; 89 seconds for a lost disk, with nothing done by hand
Backups kept7 days on Free and Plus, 14 on Pro, never fewer than two full backups
Point-in-time restorePro: any moment in the last 7 days, into a new project
Restore testingEvery project's backup restored and checked at least weekly
Deletion lockEvery backup version locked for 7 days, including against us
Second regionEvery backup replicated to us-east-1; restoring from it has been drilled

All of it is on backups and recovery.

Is it up, and has it been?​

status.snoutdata.com shows hosted databases, project APIs, creating projects, sign-in, the dashboard and the rest, checked every fifteen minutes, with 90 days of history, incidents and maintenance. It never shows green without a fresh reading.

Who can see my data?​

An operator of ours can reach a hosted database, and we say so rather than claim otherwise. Each project runs in its own rootless container, as a role that is not a superuser, with storage credentials scoped to its own prefix. Realtime, file storage and the function runtime are shared per machine. The project password is stored encrypted under a key held apart from the database. See security. If a credential that never leaves your machine is a requirement, Studio is the product for that.

Compliance​

SnoutData is not SOC 2 certified. The control set is designed to the Trust Services Criteria, with no Type II audit performed. A security overview mapped to those criteria, a subprocessor list and a data processing agreement are available on request from security@snoutdata.com.

What does it cost, and what are the limits?​

Free, Plus and Pro, with projects, storage, connections, API requests a minute, memory and CPU per plan written down on limits, along with what is not built. Plans and billing are on plans.

Which Postgres?​

Postgres 18, upstream and unmodified, with pgvector, pg_cron, pg_net, pg_graphql and SnoutTime in the image (extensions). Data checksums are on, so corruption is reported rather than returned. A new Cloud project runs 18. A project made on 17 keeps running on 17, supported like any other, with nothing for you to do; the same holds for a database on your own machine (snoutdata start, a Local project in Studio, a self-hosted stack). Postgres 18 has the detail.

Is it moving, and in which direction?​

Every user-visible change is dated in the Cloud changelog (also an Atom feed), and Studio's in the Studio changelog (feed).

Can I leave?​

Yes, with the same schemas either way:

  • Export a project from the dashboard, the CLI or Studio, and restore it anywhere that runs Postgres.
  • Run the same stack yourself with Docker Compose (self-hosting): the same servers, paths and keys, so your application changes only its URL.
  • Move a database between Cloud and any Postgres, in either direction, with a row count of every table on both sides at the end.