Service level agreement
Availability of the console and API, how it is measured, credits, recovery targets.
What it is
The availability Deutsche Cloud commits to for the console and the API, how it is measured, the credit you receive when a month falls short, and the recovery targets for the platform's own database. Live state: status.platform.deutschecloud.app.
Not in effect yet. This SLA applies to usage from a date published here and on the website's service-levels page. Availability before that date is published for information only.
Concepts
- Control plane — the console, the public API (and the CLI, Terraform and MCP access that go through it), and the Git, npm and container-registry endpoints the platform serves. This SLA covers the control plane only.
- Your resources — machines, databases, apps, buckets, queues and the rest. They keep running while the control plane is down. Their service levels are not part of this page.
- Probe — an uptime probe in Helsinki, outside the platform it watches. It requests the console's health endpoint every 30 seconds with a 10-second timeout and serves the status page.
- Unavailable minute — a minute in which the probe got no successful answer. With a second probe location, both must see the failure.
- Monthly availability — available minutes divided by all minutes of the calendar month.
Legal review: scope of the covered services; the calendar month's time zone; whether the probe's record is binding evidence.
How to
Check availability
Open the status page. It answers from the probe even when the console does not, says since when the console has not answered and from where, and shows the last state the platform published.
Claim a credit
Write to info@deutschlandgpt.de within 30 days after the end of the month, naming your organization and the month. Times and request IDs of failed requests you saw are accepted as evidence; they can add unavailable minutes, never remove them. The credit is set against your next invoice.
Legal review: the 30-day claim deadline; set-off against the next invoice only; whether credits are the sole remedy.
Reference
Commitment
| What | Value |
|---|---|
| Availability | 99.9 % per calendar month |
| Allowed downtime | about 43 minutes in a 30-day month |
| Measured by | the probe, per minute (see Concepts) |
| Effective from | not yet set |
Credits
A share of your organization's whole invoice for the month:
| Monthly availability below | Credit |
|---|---|
| 99.9 % | 10 % |
| 99.0 % | 25 % |
| 95.0 % | 50 % |
Credits are capped at 50 % of the month's invoice and are never paid out. Free and trial accounts are measured and see the same numbers, but receive no credits.
Legal review: the credit base (whole invoice) and the 50 % cap.
Exclusions
Minutes in these situations are not counted as unavailable:
- Scheduled maintenance announced on the status page at least 5 business days ahead, outside 08:00–20:00 CET, at most 2 hours a month. Deploys and planned database switchovers are not maintenance and are counted.
- Unavailability you caused: your credentials, your configuration, abuse from your account.
- Force majeure: war, natural disaster, action by a state authority.
- Features marked as preview.
An outage at the hosting provider, including the loss of a whole data-centre location, and failures of upstream networks or DNS providers are counted.
Legal review: the force-majeure definition and its exclusion of hosting-provider outages; the maintenance notice period and cap.
Recovery targets for the platform database
Targets, not credited commitments:
| Failure | Data loss (RPO) | Back in service (RTO) |
|---|---|---|
| One database machine lost | none | ≤ 1 minute |
| The whole database cluster lost | ≤ 5 minutes | ≤ 1 hour |
Legal review: the "targets, not credited" clause.
What the drills measured
On 2026-11-13, on a copy of the production layout (a load balancer, two control-plane machines in Falkenstein and Nuremberg, a three-machine PostgreSQL cluster), from outside, one request every 250 milliseconds and one write every 500 milliseconds:
| Failure | What clients saw | Writes lost |
|---|---|---|
| A control-plane machine powered off | no failed request; some answers took about 5 s for 8 s | 0 |
| The app on one machine stopped | about half the requests failed for 9 s | 0 |
| The database leader's machine powered off | every request failed for 30 s (44 s with default timing) | 0 |
| Planned database switchover | every request failed for 3 s | 0 |
| All three database machines lost, restored | 2.5 minutes, with the steps prepared | the last 4.6 minutes |
| The Falkenstein location lost | down for as long as it was, plus 40 s | 0 |
What is not supported
- No SLA for your resources (machines, databases, apps, storage) on this page.
Kubernetes clusters of the
standardtier have their own API availability SLA, measured per cluster: see Managed Kubernetes. - One probe location: until a second one runs, its view alone decides.
- Two of the platform database's three machines are in Falkenstein: losing that location stops the console, and those minutes count.
- No credits for free and trial accounts.
Troubleshooting
| Symptom | Cause | Fix |
|---|---|---|
| The status page is up but the console does not answer | The control plane is down; the probe runs outside it | The status page says since when; claim if the month falls below 99.9 % |
| The console fails for you but the status page is green | The failure is between you and the platform, or only some requests fail | Send the times and request IDs with your claim |