Service Level Agreement
What we commit to for hosted and managed services, how it is measured, and what happens when we miss. This is the summary; the signed agreement for your service governs.
Availability
Target availability for managed production services is 99.9% per calendar month, measured at the edge by synthetic probes from outside our network, excluding scheduled maintenance.
Measurement is ours and is published to you monthly. Where your own monitoring disagrees materially with ours, we investigate rather than argue.
Response times
Acknowledgement time is set by plan: Site — one business day. Application — four business hours. System — one hour, around the clock.
On every plan, production that is down or data at risk is worked continuously until it is resolved, whatever tier was bought.
The commitment is to acknowledgement and to work, not to a repair time: a fault whose cause sits with a third party does not resolve on our word.
Maintenance
Routine maintenance is performed in a window agreed with you and announced in advance. Security patches that cannot wait are applied immediately and reported afterwards.
Deploys to production are zero-downtime and reversible.
Backups and recovery
Backups run every night, automatically. Seven copies are kept for each site and each database, up to thirty days, and the newest copy of a target is never deleted.
Every night one of them is actually unpacked and checked for content — not merely checked for a valid file. A backup that has never been restored is not counted as a backup, which is why that check runs nightly rather than quarterly.
One disclosure: backups are currently kept on the same machine they came from, and are not encrypted at rest. An off-machine copy and encryption are the next step, and are not claimed here before they exist.
When we miss
If we fall below the availability target in a month, you receive a service credit against the following month, calculated as set out in your agreement.
You also receive a written account of what happened, why, and what changed so that it does not happen again. The credit is the smaller part of that.
Exclusions
Downtime caused by your own code changes deployed outside our pipeline, by third-party services outside our control, or by force majeure, is excluded from the availability calculation.
We will still work the incident. It just does not count against the number.