Plan and manage an Azure AI solution 1.2 Set up AI solutions in Foundry
A company is designing a Foundry-based solution that must keep working if one Azure…
Study guide skill: Design Azure infrastructure for AI apps and agent-based solutions
Question B6-005
A company is designing a Foundry-based solution that must keep working if one Azure region becomes unavailable. The solution also calls the Translator API. Which two design decisions are correct? (Choose two.)
Select 2 answers.
Answer: A, D
Foundry doesn't support automatic cross-region failover, so multi-region availability means separate Foundry resources per region, with data synchronization and routing handled in the application layer. Most new APIs are available at the project scope, but some, such as the Translator API, are available only at the Foundry resource level. Plan the deployment structure around the API scopes the workload needs.
Deep dive
How it works
Two separate facts from the Foundry architecture page answer this question.
Regional resilience is your job. Foundry doesn't support automatic cross-region failover. If you need multi-region availability, you deploy separate Foundry resources in each target region and manage data synchronization and routing in your application layer. There's no built-in replication you can switch on. Before you provision, confirm that each target region supports the capabilities you need.
API scope matters. Most new APIs are available at the project scope. Some capabilities that Azure OpenAI, Speech, Vision and Language originally offered at the account level are available only at the Foundry resource level, not at the project scope. The page gives the Translator API as the example. So calls to Translator target the Foundry resource, not a project, in each region.
Put together: one Foundry resource per region, application-level routing and sync, and Translator calls aimed at the resource rather than a project.
The other options, and when you'd use them
- One resource with projects replicated to a paired region: There's no automatic failover to rely on. Microsoft's answer for multi-region is separate resources per region, with sync handled by you.
- Enable automatic cross-region failover: That setting doesn't exist. Foundry doesn't support automatic cross-region failover.
- Call Translator through each project's endpoint: Not every API is at project scope. Translator is available only from the Foundry resource level.
Exam tip
Foundry multi-region = separate resources + your own routing and sync. Translator = resource level, not project scope.
References
Last reviewed: