Question

A software company is launching a new multi-tenant SaaS analytics platform. Each tenant generates unpredictable, spiky query loads, and the number of tenants is expected to grow from dozens to thousands over 18 months. Tenants require strong data isolation for compliance, and the platform must scale storage and compute independently while minimizing idle capacity costs during off-peak hours. The team wants a relational data store that supports SQL analytics with automatic scaling and near-zero administrative overhead. Which design best meets these requirements?

Accepted Answer

Use Amazon Aurora Serverless v2 with PostgreSQL compatibility, isolating tenants across separate databases/schemas, so compute (ACUs) scales automatically with load while storage grows independently and idle capacity shrinks.

Answer

Provision a single large Amazon RDS for PostgreSQL Multi-AZ instance and use PostgreSQL schemas to separate each tenant, scaling the instance class up as tenant count grows.

Answer

Deploy a fixed fleet of Amazon EC2 instances running PostgreSQL behind a Network Load Balancer, and add read replicas manually as tenant demand increases.

Answer

Store all tenant data in Amazon DynamoDB with a tenant-ID partition key and run analytics queries using PartiQL, relying on on-demand capacity mode for elasticity.