How Many Tenants Can You Handle Before Performance Degrades?

SaaS capacity planning requires a different load model from single-tenant applications: the relevant metric is not just total concurrent users but concurrent tenant activity patterns, because different tenants may share infrastructure while having uncorrelated load patterns. Peak load for a SaaS platform occurs when multiple large tenants generate concurrent activity — which may happen at specific times (business hours in different time zones) or during tenant-specific events (end-of-month processing).

The most important capacity question for a multi-tenant SaaS is how shared infrastructure behaves under simultaneous tenant load spikes. A database connection pool that comfortably serves 50 small tenants may saturate when 3 large enterprise tenants simultaneously run intensive data exports. We model this scenario by simulating tenant-specific load profiles — including the bursty patterns that enterprise tenants create — and measuring the point at which shared infrastructure becomes the binding constraint.

Enterprise tenant onboarding capacity is a specific scenario that SaaS platforms frequently undertest: adding a single large enterprise tenant can change the load profile of shared infrastructure significantly. We simulate the impact of onboarding a tenant with 10x, 50x, and 100x the data volume and activity of your typical tenant, and identify the infrastructure changes required to support enterprise growth without degrading existing tenants.

Key Challenges for SaaS Platforms

Multi-Tenant Load Simulation — Designing load tests that model realistic concurrent tenant activity, including enterprise tenant data volumes and bursty usage patterns.

Shared Infrastructure Limits — Identifying database connection pool saturation, cache eviction pressure, and message queue backlog growth under simultaneous multi-tenant load.

Enterprise Tenant Capacity Impact — Measuring the infrastructure impact of adding a single large enterprise tenant and determining the scaling requirements before onboarding.

Auto-Scaling Validation — Testing whether auto-scaling responds fast enough for the bursty load patterns that SaaS platforms experience, including simultaneous spikes from multiple tenants.

Cross-Portfolio Resources

SaaS platforms also need: performance.qa for multi-tenant query performance optimisation and API P99 improvement, and stresstest.qa for shared infrastructure resilience testing and chaos engineering.