For high availability of OCI Functions executions across availability domains, what subnet choice does official guidance prefer when available?

For high availability of OCI Functions executions across availability domains, official guidance prefers a regional subnet when one exists so executions can fail across ADs without listing one subnet per AD. HA for Functions at this layer is an application subnet choice: wider regional placement reduces correlated AD loss for civic workloads without inventing a second Functions product.

Using one AD-specific subnet only and never listing more than one fails by concept when multi-AD resilience is the goal—that choice pins executions to a single AD’s subnet. Peering subnets in a second region for every function invoke fails because cross-region peering for each invoke is not the documented in-region HA subnet pattern and confuses DR topology with application subnet selection. Claiming no subnet at all when the application uses detached invoke fails because detached versus sync changes invoke semantics, not the requirement to attach the application to subnets.

Exam caveat: if only AD-specific subnets exist and you need that resilience, you may name up to three—regional is preferred when available, not the only legal shape. Operational check: inspect the Functions application’s subnet list, prefer a regional subnet in the municipal VCN, invoke under failure-injection or AD maintenance windows in non-prod, and confirm executions continue on remaining capacity.