A multi-tenant analytics vendor keeps one database per client. Usage across clients is sporadic and peaks land at unrelated hours. The vendor wants to keep strict data separation and control spend. Which approach fits best?
Show answer and explanation
Answer: B
An elastic pool lets many single databases draw from a shared budget of compute and storage. Because each tenant's peaks occur at different times, the pool absorbs them without every database needing capacity for its own peak. Each tenant keeps its own database, so data separation stays intact while cost falls.
Why the other options are wrong
- A. A VM per tenant gives full control but multiplies OS, patching and licensing overhead. Idle VMs stay billed for compute unless someone deallocates them, which makes this costly for sporadic usage.
- C. A Managed Instance per tenant is heavyweight and billed per instance. Paying for an instance for each idle client wastes money when a pool can share capacity across them.
- D. Individually provisioned databases must each be sized for their own peaks, so idle capacity is paid for repeatedly. That is the cost pattern pooling is meant to avoid.