SLA Specification and Negotiation Model for a Network of Federated Clouds: CloudLend

The Cloud computing paradigm is remarkably evolving. Cloud customers have become more conscious about the QoS expectation from Cloud providers' services. Accordingly, Cloud providers are required to be responsive to customers' demands and be open to establishing federations with other providers in order to retain their shares in the competitive Cloud market. Cloud customers are always searching for optimized services, irrespective of which providers are taking part in a federation to deliver these services. They seek federated Cloud services to attain the maximum satisfaction level, which is measured by the degree of adherence to customers' quality of service (QoS). Such federations of Cloud services are typically governed by service level agreements (SLAs) that are negotiated between the Cloud customer and provider prior to service provisioning. This paper tackles the challenges related to SLA specification and negotiation in a federated network of Clouds, CloudLend. We first propose a weighted SLA specification model that captures customers' QoS and manages multi-level SLAs specification. We then introduce an autonomous SLA negotiation model that adopts an enhanced fair division game. The model enables federated Cloud services to examine SLAs, react to SLA offers and eventually sign an SLA contract. It autonomously detects changes in Clouds federations and revises SLA specifications accordingly. The proposed model is evaluated using a CloudLend simulator, which we developed for this purpose. Several test cases were executed, and the results we have achieved verified the fairness and efficiency of our proposed SLA specification and negotiation models in CloudLend.