Optimizing the deployment of cloud-hosted application components for guaranteeing multitenancy isolation

Abstract

The sharing of components of a cloud-hosted application and the underlying cloud resources between tenants have the potential to reduce resource consumption and running cost per tenant. However, this sharing may allow the behaviour of one component to affect the performance, resource consumption and access privileges of other components, for example, if the application fails to scale-up when the workload of one of the component increases suddenly. This problem becomes even more acute when components are associated with different or higher degrees of isolation among them. This paper presents a mathematical optimization model and a metaheuristic (based on simulated annealing) solution for providing near-optimal solutions for deploying components of a cloud-hosted application in a way that guarantees the required degree of isolation between the components. Our experiments showed that the near-optimal solutions obtained from our model had low variability and percent deviation, when compared with the optimal solution. We also provide some recommendations in deploying components associated with varying degrees of isolation.