Modern Software-as-a-Service (SaaS) applications increasingly adopt multi-tenant architectures to optimize resource utilization, scalability, and operational efficiency. This thesis evaluates how different tenant isolation strategies impact performance and infrastructure costs in monolithic and microservice multi-tenant architectures. The study uses Deskly, a multi-tenant resource reservation system, implemented in both architectural styles. Each version supports three levels of tenant persistence isolation, including row-level, schema-level, and database-level, and was deployed on the AWS cloud infrastructure. The performance characteristics of each architecture were assessed in the multi-tenant specific scenario like tenant onboarding as well as business ones like resource reservation using load and stress tests. The findings offer insights into architectural trade-offs and financial implications, helping SaaS developers make informed decisions.
Kamil Bońkowski (2026) studied this question.