A Hybrid Multi-Tenant Database Schema for Multi-Level Quality of Service

Abstract

Software as a Service (SaaS) providers can serve hundreds of thousands of customers using sharable resources to reduce costs. Multi-tenancy architecture allows SaaS providers to run a single application and a database instance, which support multiple tenants with various business needs and priorities. Until now, the database management systems (DBMSs) have not had the notion of multi-tenancy, and they have not been equipped to handle customization or scalability requirements, that are typical in multi-tenant applications. The multi-tenant database performance should adapt to tenants workloads and fit their special requirements. In this paper, we propose a new multi-tenant database schema design approach, that adapts to multi-tenant application requirements, in addition to tenants needs of data security, isolation, queries performance and response time speed. Our proposed methodology provides a trade-off between the performance and the storage space. This proposal caters for the diversity in tenants via defining multi-level quality of service for the different types of tenants, depending on tenant rate and system rate. The proposal presents a new technique to distribute data in a multi-tenant database horizontally to a set of allotment tables using an isolation point, and vertically to a set of extension tables using a merger point. Finally, we present a prototype implementation of our method using a real-world case study, showing that the proposed solution can achieve high scalability and increase performance for tenants who need speedy performance and economize storage space for tenants who do not have demanding quality of service.