The rapid growth of software as a service (SaaS) has necessitated the design of an architectures that can simultaneously ensure scalability, security, and performance, as well as accommodate multiple tenants. Traditional multi-tenant SaaS systems continue to have problems with workload isolation, where sharing of resources among tenants can lead to performance variability and undermine SLA. This paper presents a next-generation multi-tenant SaaS system with the assistance of AI-driven resource isolation. We propose to apply the dynamic scaling mode, which presents a scalable solution to the problem of dynamic workload prediction, resource allocation, and enforcement of isolation policies regarding tenant interference. The method not only increases scalability but also provides predictable performance across heterogeneous workloads, a feature that is largely absent from most current solutions. Experimental results and comparative studies to baseline models indicate that the proposed approach can substantially improve throughput and reduce latency as well as tenant-level quality of service (QoS). The study has a bearing on the future development of SaaS deployment since it outlines how the multi-tenancy model can be optimized through the orchestration of workloads by AI to be more efficient, secure, and scalable.
Ravi Chandra Thota (Sat,) studied this question.