Data Residency
also called Data Sovereignty
A requirement that specific data be stored and sometimes processed only within a defined geography.
This is the constraint that most often forces an architecture into multiple regions for reasons that have nothing to do with resilience, and it needs to be established early because retrofitting it is close to a rebuild.
The questions to settle precisely, because they have very different costs: does the requirement cover storage only, or processing too? Do backups and disaster-recovery copies count (they do)? Do logs and telemetry count (they usually do, and are usually forgotten — a trace payload or an error log can carry personal data to a central observability region)? Does a support engineer viewing a record from another country constitute a transfer?
Common architecture: regional data planes holding the regulated data, with a global control plane holding only non-regulated metadata, plus identity federation so a user is routed to their home region. The hard parts are cross-region features — global search, aggregate reporting — which have to be satisfied with anonymised or aggregated data.