Cost-Efficient Utilization of Public SLA Templates in Autonomic Cloud Markets

Abstract

Dynamic and self-adaptable markets are fundamental for the successful implementation of the Cloud computing paradigm where computing resources are provided on demand and dynamically to the heterogeneous user base. Usually, in Cloud markets, contracts between traders are established using Service Level Agreements (SLAs), which include objectives of service usage. However, Cloud markets face challenges, which are usually not evident in other market types, such as non-standardized goods and large variety of resource types offered on the market, resulting in low number of matches of consumers' bids and providers' asks. Moreover, existing Cloud market mechanisms are usually static and cannot react on dynamic change of user requirements. To counteract these problems, we propose an adaptive approach for autonomically deriving public SLA templates (i.e., templates for generation of electronic contracts) based on user requirements. Although this approach brings many benefits, it also incurs cost to users, since they must adapt specifications of their requirements in order to match automatically generated public SLA templates. In this paper, we reduce this cost by autonomically modifying specifications of user requirements. We introduce a method for adapting public SLA templates not only in respect to definitions of SLA attributes, but also to attribute values. Based on an appropriate utility and cost model, we evaluate the approach and show that it significantly improves the performance of the adaptive SLA template technique.