Service Level Agreement

Forge Metal is currently deployed on a single bare-metal node. Every component — the compute orchestrator, databases, identity, billing, ingress — runs on that node and is not replicated across failure domains. A hardware failure, a kernel crash, or a datacenter-level event is therefore a correlated outage of the whole service.

We do not offer a guaranteed availability percentage or service credits on this tier. This is not a gap in the product; it is a property of single-node deployment, and pricing reflects it. Customers whose workload needs an availability SLA should wait for the three-node tier described below or bring their own redundancy.

The founder monitors the node with the same observability stack described in the Security Overview. Material incidents are posted on the changelog and mirrored into each affected organization's audit trail.

Planned maintenance that may affect availability is announced at least 48 hours in advance to account administrators and is performed in a published maintenance window. Emergency maintenance (security patches, active-incident response) is performed without prior notice; a post-incident writeup is published on the changelog.

Support response and resolution targets are defined in the support plan named on your order form. In the absence of a specified plan, support operates on best-effort, business-hour basis through the policy mailbox.

Security reports take precedence over routine support correspondence; they route through the security mailbox and are acknowledged on receipt. See Coordinated disclosure.

The next topology introduces three nodes with cross-node replication for every stateful component — consensus replication for the ledger, replicated analytics storage, streaming replication for the relational store, and send-based replication for durable customer volumes — all carried over a private overlay network between nodes. With this topology comes a 99.9%-per-calendar-month availability commitment, measured as the percentage of minutes in which each customer-facing service endpoint responds with a non-5xx status to a synthetic probe.

Downtime below 99.9% triggers service credits on a sliding scale against the affected month's subscription fees. The exact credit schedule, exclusions (force majeure, customer-caused outages, dependencies on third-party systems), and claim procedure will be published alongside the three-node tier's general availability.

Material changes take effect 30 days after they are announced by email to the administrators on each affected organization. The effective date at the top of this page is the date the current version took effect. Prior versions of all policies live at /policy/changelog, and every change is recorded there in commit-addressable form.

Policy identifier: sla.

Questions and requests under this policy go to policy@anveio.com. Security reports go to security@anveio.com and take precedence over routine policy correspondence. GDPR data-protection correspondence may be directed to dpo@anveio.com; abuse reports to abuse@anveio.com.