Multi-Region Web State and Recovery

Place work across regions without hiding read staleness, write ownership, or recovery loss.

A second region changes more than request latency. Public assets, private responses, write authority, replica freshness, queue consumers, backups, and telemetry may each live in a different place. This track follows a case-review service whose readers work from two regions while one region owns writes. It covers routing and cache scope, read-after-write behavior, failover fencing, and data-location inventories. The examples are admission decisions, not provider-specific deployment scripts; actual guarantees depend on the chosen storage and network services.

Topics in this track

Regional Routing, Static and Private Cache Boundaries — Serve public assets broadly while keeping authenticated responses tied to their correct owner and region.
Replica Lag, Read-Your-Write, and Version Cursors — Give recent writes a freshness path instead of pretending all regional replicas are current.
Regional Failover, Fencing, and Replay — Promote a recovery writer only after old write authority is contained and uncertain work is reconciled.
Regional Data Location and Operational Evidence — Inventory every place private data, logs, backups, and derived artifacts can reside.