Towards Self-Adaptive Cloud Collaborations

In multi-tenant collaborative cloud systems, the choice of cloud providers is based on pre-negotiated terms and conditions to help meet Service Level Agreements (SLAs) and streamline resource provisioning. However, these collaborations are not self-adaptive to accommodate changes in tenancy requirements, quality of resources and provider capabilities as provider collaborations remain static over tenancy period. In our paper, we present an approach towards achieving self-adaptive cloud collaborations. Our model selects the best possible cloud collaborations from amongst multiple possible dependent cloud provider collaboration links. The formation of cloud provider collaborations is broadly based on cloud provider capabilities, tenancy requirements, cost modeling of each cloud provider andits functional ability at the Software as a Service (SaaS) layer. We present an algorithm to form dynamic cloud collaborations and thereby determine the most appropriate linkages within the providers. We also illustrate our approach throughout our paper with a realistic running example, and present detailed experimental evaluation via a proof-of-concept prototype implementation.