This document describes HAIEC's business continuity and disaster recovery posture, including infrastructure redundancy, backup strategy, recovery objectives, and incident escalation procedures.

1. Infrastructure Redundancy

Vercel: Multi-AZ within US-East region. Serverless functions automatically distributed across availability zones.
Neon: Serverless PostgreSQL with point-in-time recovery (PITR). Data stored in US-East.
Upstash: Multi-zone Redis replication within region.
Modal: Ephemeral containers. No persistent state to recover.

Honest gap: HAIEC operates in a single region (US-East). We do not have multi-region failover. A region-wide outage would affect platform availability until the provider resolves it. Multi-region deployment is a roadmap item.

2. Backup Strategy

Database: Neon point-in-time recovery (PITR). Allows restoration to any point within the retention window.
Application code: Version-controlled in GitHub. Full history available.
Configuration: Environment variables stored in Vercel dashboard. Exportable.
Evidence records: SHA-256 content hashes provide tamper detection. Original content stored in Neon.

Honest gap: We do not currently perform automated full database backup exports to a separate storage provider. We rely on Neon's PITR capability. Automated backup exports to cold storage are a roadmap item.

3. Recovery Objectives

Metric	Target	Status
RTO (Recovery Time Objective)	4 hours	Target — not formally tested
RPO (Recovery Point Objective)	1 hour	Target — depends on Neon PITR window
Application rollback	Seconds	Achieved via Vercel instant rollback
Database PITR	Within Neon retention window	Available via Neon

Honest gap: RTO and RPO targets have not been formally tested through disaster recovery drills. We are working toward formal DR testing as part of our SOC 2 readiness program.

4. Disaster Scenarios & Response

Vercel outage (application unavailable)

Wait for Vercel to resolve. Monitor Vercel status page. Notify customers via email if outage exceeds 15 minutes. No manual failover possible — platform is fully dependent on Vercel.

Neon outage (database unavailable)

Application will return 503 errors for database-dependent routes. When Neon recovers, use PITR to restore to last known good state if data corruption occurred. Static pages and cached content remain available.

Data corruption

Use Neon PITR to restore database to point before corruption. Reconcile any data written between PITR point and incident detection. SHA-256 evidence hashes help verify integrity of restored records.

Region failure (US-East)

No automatic failover. Platform unavailable until region recovers. Manual redeployment to alternate region possible but requires configuration changes. This is a known limitation.

5. Incident Escalation

Detection: Sentry alerts (application errors), Vercel monitoring (infrastructure), rate limiting alerts (abuse), customer reports via security@haiec.com.
Escalation path: Alert → security@haiec.com → founder/engineer on call.
Customer communication: Email to affected customers. Status updates as investigation progresses.
Regulatory notification: GDPR breaches reported to supervisory authority within 72 hours (Article 33).

See Incident Response Plan for the full 6-phase response process.

Business Continuity & Disaster Recovery