Multi-Site Replication

The general architectural pattern of copying or synchronizing S3-compatible object data across two or more geographically distinct storage locations. Encompasses several specific shapes — primary→secondary one-way sync (most common, lowest complexity), bidirectional / active-active (full read-write at every site), and edge→core aggregation (many sites feeding a single central lake). The variant choice is driven by RPO/RTO requirements, write-locality needs, and the tolerance for conflict-resolution complexity.

Single-site object storage carries three structural risks — provider outage, regional disaster, and vendor lock-in. Multi-site replication mitigates all three by maintaining a redundant copy somewhere the primary failure mode cannot reach. The replication topology is the load-bearing decision: one-way sync is operationally simple but adds failover latency; active-active eliminates failover but introduces conflict resolution; edge-to-core scales out for IoT-shaped workloads but is unidirectional. Tools like **rclone** and provider-native replication (S3 Cross-Region Replication, GCS Storage Transfer, MinIO mirror) implement the data-plane mechanics; the architecture node names the pattern.

Disaster recovery with bounded RPO, geographic data residency compliance, multi-cloud architectures for vendor-independence, multi-region read acceleration via local replicas, cross-provider migration with continuous sync until cutover.