A client says "our data must stay in country". What do you ask before designing anything?
Show the full answer Hide the answer
What the interviewer is testing
Whether you interrogate a compliance requirement to find its actual scope, since the answers differ by orders of magnitude in cost.
The questions
Which data? All of it, or personal data, or a specific regulated category? Frequently only one category is constrained, and the rest can be global.
What operation is constrained? - Storage at rest in country — the easiest - Processing in country — rules out a global control plane touching the data - Access from within the country — excludes support engineers elsewhere, which is an operating model change rather than a technical one
What is the legal source? A statutory requirement, a regulator's guidance, a contractual clause, or a procurement preference. These have very different negotiability, and a surprising proportion of "requirements" turn out to be the last.
Does it cover metadata? This is the clause that catches teams out. Backups, logs, telemetry, search indexes, monitoring data and disaster recovery copies are data too, and they default to elsewhere. Establish this explicitly.
Where must disaster recovery be? If failover must also be in country, a second in-region site is needed rather than the natural cross-region alternative.
Why the answers matter so much
| Scope | Consequence |
|---|---|
| Personal data at rest only | Regional storage; global services can process metadata |
| All data, processing included | Full regional stack; shared services need regional deployments |
| Plus access restriction | Regional support staffing and access controls |
| Plus in-country DR | Two regional sites, doubling the footprint |
The architectural warning
Residency is a partitioning decision, not a deployment parameter. It touches identity, routing, data model, aggregation and operations. Retrofitting it into a globally replicated system is close to a rebuild, which is why it must be established before the data layer is designed.
What a strong answer adds
Global aggregation as the hard part: reporting across regions on data that cannot leave requires computing aggregates locally and combining only the results, which constrains what analytics is possible. Establishing that constraint early prevents a reporting requirement being promised that the architecture cannot deliver.
Common weak answers
Selecting a region and declaring compliance. Treating it as an infrastructure decision.