Architectural models of multitenancy in modern SaaS systems: a comparative analysis and selection methodology

Abstract

In the modern software industry, where the "Software as a Service" (SaaS) model is highly dominant, multitenancy has become a fundamental architectural principle for achieving cost-effectiveness and scalability. The selection of an optimal multitenancy model is one of the most critical strategic decisions, directly influencing the system's cost, security, performance, operational complexity, and overall capability. An incorrect choice can lead to severe consequences, such as potential data leakage between tenants, performance degradation due to the "noisy neighbor" problem, or the creation of an unsustainable business model altogether. Existing scientific literature provides detailed descriptions of classic models (Silo, Pool); however, it often considers them within the context of monolithic applications, which is insufficient given current software development trends. With the increasing complexity of software products, a need has emerged for a more flexible approach, where different components within a single application may require different strategies depending on their functional and non-functional requirements. This research offers a systematic comparative analysis of the primary architectural models of multitenancy: from complete tenant isolation (the Silo model) to various types of resource sharing ("Database-per-tenant," "Schema-per-tenant," "Shared Schema for all tenants") and other hybrid approaches. The analysis is conducted across many significant criteria, including data isolation, cost per tenant, scalability, performance, development complexity, and the potential for personalization. The scientific novelty of this work lies in the development of a well-founded methodology for selecting the optimal strategy, presented as a structured set of criteria well-suited for decision-making. These criteria help software architects align technical solutions with business requirements, such as the target market, security regulations (GDPR, HIPAA), pricing models, and other demands