AI-Enabled Architecture for Large-Scale Multi-Tenant Systems

Abstract

A Software-as-a-Service provision delivers pre-configured solutions for its clients. Tenants with varying quality requirements are assigned to distinct dedicated instances. The issues associated with producing inconsistent quality responses from a single instance are not explicitly addressed in current design methodologies. The absence of standardized strategies and design protocols complicates an architect's ability to incorporate multi-tenant design choices during the initial phase. This work emphasizes several domain-independent architectural considerations for managing multiple heterogeneous tenants on a shared application instance. We identify essential quality requirements pertinent to the multi-tenant scenario, associated tactics, metrics, and assess their influence on other software product quality attributes.