Availability

Last updated: 2026-04-11 • ← Infrastructure

Service level agreement

Backbuild targets a monthly uptime of 99.9% for the platform. Formal SLA terms, including the precise uptime definition and measurement methodology, will be published in the Terms of Service. Incidents that cause the monthly target to be missed will be eligible for service credits in accordance with those terms.

Status page

The public status page displays current service status, component health, uptime metrics, incident history, and scheduled maintenance windows. The page checks the API health endpoint in real time and auto-refreshes every 60 seconds. Incident notifications are also sent directly to designated customer contacts.

Redundancy

• Edge compute: Cloudflare Workers run across hundreds of points of presence worldwide. Failure of any individual location is handled automatically by Cloudflare's network: traffic routes to the next nearest location with no customer action required.
• Database: the managed Citus PostgreSQL cluster is configured with replication so that loss of any single node does not interrupt service. Write availability is maintained through coordinator failover.
• Object storage: Cloudflare R2 provides durable, replicated object storage with high availability.
• DNS and TLS: DNS and certificate services are provided by Cloudflare's global, redundant infrastructure.

Backups

• Full backups: the primary database is backed up daily. Each backup is verified with a SHA-256 checksum, encrypted at rest, and stored in object storage.
• Write-ahead log shipping: WAL shipping is operational and produces hourly incremental backups that support point-in-time recovery.
• Off-site storage: backups are stored in object storage that is geographically separated from the primary database.
• Retention: backups are retained per the backup rotation policy; historical backups age out on a documented schedule.

Recovery objectives

Objective	Target
RTO — API tier	4 hours
RTO — Database coordinator	4 hours
RTO — Full database cluster rebuild	8 hours
Recovery Point Objective (RPO)	1 hour

These targets reflect the stated design objectives of the platform. They will be confirmed through formal disaster recovery testing as part of the business continuity program. See business continuity for details on DR testing and the BCP process.

Maintenance windows

Routine maintenance is performed in a rolling, zero-downtime manner wherever possible. Disruptive maintenance is scheduled in advance and customers are notified through their designated communication channels. Emergency maintenance may occur without advance notice when required to address security or reliability issues.