The gap between "backed up" and "still running."
Most multi-location businesses think "we have backups" means "we have disaster recovery." It doesn't. Backups answer the question can we get the data back? — eventually, after a restore, often with hours of lost work. Live replication answers a different question: if one office is gone, can the others keep working today?
That second question matters when a Saturday-night fire hits the practice's primary server room and thirty staff at the other locations are scheduled to start work Monday morning. Backups will eventually get the data back. They won't make Monday morning happen.
What we built.
A multi-site replication architecture that continuously syncs critical business data across every office. The data at every site is current within a few minutes of changes anywhere in the company. When a site goes down, switching the surviving offices over to the replicated copy is a documented, rehearsed procedure — not a panic.
Replication Layers
- Block-level or file-level replication tuned to each business system: file servers, SQL databases, business application data, email archives, document management stores
- SQL Server AlwaysOn Availability Groups for transactional databases
- MySQL and PostgreSQL streaming replication for open-source database workloads
- DFS-R, Resilio Connect, or custom rsync pipelines for file-server replication
- NAS-to-NAS sync for Synology, QNAP, NetApp, and similar storage platforms
- Application-aware replication for systems with their own native multi-site features
Recovery Targets
- RPO (recovery point objective) measured in minutes — not days
- RTO (recovery time objective) measured in hours — documented switchover, not a guess
- Quarterly tested failovers in a controlled window: we actually run the switchover so we know it works on the day you need it
Operations
- Bandwidth-aware: replication piggybacks on the existing SD-WAN or site-to-site VPN with QoS so business traffic stays fast
- Encryption in transit (TLS 1.3) and at rest (AES-256)
- Integration with the existing backup architecture: replication handles "the lights stay on" continuity; immutable cloud backups still cover the ransomware-hits-every-site case
- Documented runbooks for the most likely failure modes — failover, failback, partial-replication catch-up
What it demonstrates: We treat disaster recovery as an engineering problem with measurable targets, not a compliance checkbox. The deliverable is a tested, documented switchover procedure with RPO and RTO numbers your insurance underwriter will recognize — not just a vendor invoice and a hope that the backup tape is readable.