Service levels & availability

Permafrost is built for high availability and we operate it to that standard on a best-effort basis. We do not publish a contractual availability percentage at the current tier, and we will not put a number in front of you that we cannot stand behind. When we commit a contractual figure, it will appear here and in your agreement, not in marketing copy.

Enterprise agreements can include committed availability and support terms. If your procurement process needs a contractual SLA, raise it with your account team and we will scope it.

Permafrost reads your connected tenants on a schedule; your posture is a point-in-time picture from the last successful read. That design has a reassuring consequence during a Permafrost outage: your tenant's actual state does not change because our service is degraded. An outage delays the next read, it does not alter a single permission, role assignment, or finding in your environment.

When the platform recovers, the next scheduled read brings your posture current. Reads run on a recurring cadence with a minimum interval between runs, so a brief interruption typically resolves on the following cycle with no action from you. Findings reflect the last successful read until then, and the dashboard shows you when that read happened.

A degraded state means part of the platform is slow or partially impaired — for example, scheduled reads running behind. Analysis stays available; freshness may lag. An outage means a core component is unavailable. In both cases the public status page is the source of truth, and your stored posture remains intact.

We publish every incident on our status page with an honest timeline — what is affected, what we are doing, and when it is resolved. Silence on the status page means there is nothing to report, not that we are hiding it. For incidents involving a personal-data breach, our separate breach-notification commitment governs how and when we contact your account administrator directly.

We aim to acknowledge support requests within one business day, and we prioritize availability-affecting issues ahead of routine questions.