Your SaaS platform hosts 200 small tenant databases. Each tenant experiences brief, unpredictable spikes in activity, while aggregate utilization remains moderate. You want to minimize compute cost, use reserved capacity discounts, and avoid managing per-database scaling. Which deployment should you choose?

The correct option is to create a vCore-based Azure SQL Database elastic pool (General Purpose) and place all tenant databases in it, configuring per-database min/max settings. This approach is ideal for multi-tenant SaaS applications with unpredictable, spiky workloads, as it allows databases to share a common pool of resources, optimizing cost. Reserved capacity discounts can be applied to the elastic pool. Per-database min/max settings prevent a single tenant from monopolizing resources while ensuring baseline performance. Deploying a single Azure SQL Database per tenant with serverless is costly due to 200 separate databases incurring individual billing and less efficient resource sharing, even with auto-pause. An Azure SQL Managed Instance is overkill and more expensive for this scenario, designed for lift-and-shift of on-premises SQL Server instances. DTU-based Premium databases are less flexible and cost-effective than vCore elastic pools for this many databases. Azure SQL Database Hyperscale is for extremely large single databases, not 200 small ones, and sharding into schemas within one database complicates tenant isolation and management.