State the requirement
Separate requirements for provider choice, data location and recovery. They may lead to different designs. Identify which workloads need to move, which data must remain in a defined location and who confirms the organisation's obligations. Do not assume that distributing components satisfies a requirement without checking the complete data flow.
Follow identity and data through recovery
A recovery environment needs usable access, current data and the application dependencies. Test what happens when the primary provider cannot be reached, including the services used to authenticate or transfer data. A deployment that starts successfully may still fail to complete the customer's task.
Account for the operating work
Record who maintains each environment, checks drift and updates the recovery procedure. Compare the recurring effort with the requirement the design serves. The useful architecture is the one the organisation can verify and operate, with documented limits, rather than the one with the greatest number of providers.
Further reading from the original publisher: Google Cloud: hybrid and multicloud patterns ↗