A Hybrid Approach to Placement of Tenants for Service-Based Multi-tenant SaaS Application

In a service-based multi-tenant SaaS application, the number of servers on which Web service instances are deployed are limited, and tenants share the same application and services. With the purpose of lowering cost of ownership by high economies of scale, we must solve the problem that how to optimally place tenants with end users to maximize the total number of tenants without violating their Service Level Agreement (SLA). This paper proposes a hybrid approach to solve placement of tenants which is called Tenant Placement Strategy (TPS). The TPS uses a combination of resource consumption estimation model, service selection with genetic algorithm (GA), case-based reasoning (CBR) and heuristic approach. CBR is proposed for matching existing execution plans which are generated by GA. In order to fully use all types of resources of the servers, a heuristic approach is proposed for selecting the optimal execution plan based on the distance of the tenant resources consumption vector and the server residual resource vector. The results of simulated experiments show that the strategy proposed in this paper is effective in placing tenants.