A Framework for Optimized Distribution of Tenants in Cloud Applications

To be successful a cloud service provider has to target a preferably large customer group to leverage economies of scale. Therefore an application offered as a service in the cloud is often configurable regarding non-functional qualities, such as location or availability. Since many of these qualities depend on the resources on which the service is hosted, a large number of computing environments has to be managed by the service provider. This paper analyses the challenges arising from such a scenario and identifies several optimization opportunities originating from an intelligent distribution of users among the functionally equal resources with different quality of services. A framework enabling the development of distribution strategies exploiting these opportunities is defined. It allows modeling of resources, their deployment dependencies, and users with specific demands. An architecture and prototype of a management system is introduced to handle the required resource provisioning and user request routing. Several optimization strategies are defined and their performance is evaluated using statistical data of an existing cloud service provider.