Service levels

What we commit to

We operate the platform continuously and treat availability as a design constraint rather than a marketing number. Hosts are patched on a schedule, capacity is managed with headroom, backups run on the data tier independently of the workloads that use it, and configuration is held in repositories so a service can be rebuilt rather than reconstructed from memory.

Maintenance

Planned maintenance is announced in advance by email to the addresses on affected accounts. Where a change carries risk to a specific service, we agree the window with you rather than announcing it. Emergency maintenance, meaning a security fix or a fault that is actively degrading service, is applied immediately and communicated as soon as the work is stable.

Uptime credits

We do not sell financially backed uptime credits, and this is a deliberate commercial position rather than an omission. A credit scheme prices an outage at a fraction of a monthly fee, which is almost never what an outage actually costs you, and it turns incident response into a billing negotiation. We would rather spend that effort on restoring service and explaining what happened.

Where an outage is our fault, we make it right directly: service time, extended terms, engineering effort to remove the cause, or a refund of the affected period where that is the proportionate answer. If you require a contractual availability percentage with financial remedies, tell us during scoping. For some workloads we can write one into an engineering engagement; for most catalog tiers we will decline rather than promise something we have not instrumented to measure.

Incident communication

During an incident affecting your service you get a direct message from the person working on it, not a status page platitude. After an incident of any substance you get a written account of what failed, what we did, and what changes as a result.

Backups are a safety net, not your archive

Backup schedules and retention windows are published per tier and we perform restores on request. They exist to recover from our failures and your accidents. They are not a substitute for keeping your own copy of anything you cannot afford to lose: source code, world saves, media libraries, and exported data should also live somewhere we do not control.

Scope

This page describes catalog services. Engineering engagements carry their own written scope, including any response expectations agreed for a retainer.