A Model for Multi-levels SLA Monitoring in Federated Cloud Environment

摘要

Cloud providers nowadays are required to form a federation of Clouds to remain competitive and gain a share in the Cloud service provision market. This provision is governed by a mutual contract known as Service Level Agreement (SLA), to which both clients and providers are confined. Monitoring activities should be undertaken to guarantee the pre-agreed SLA. However, monitoring multiple SLA in a federated Cloud depends highly on the collaboration among Cloud providers participating in satisfying a client service request, which makes it very challenging and complex. Monitoring linked SLAs requires a consistent specification of SLA parameters, dynamic SLA negotiation, and multi-level SLAs monitoring, in addition to a reliable enforcement measures. In this paper, we extend SLA specification to cope with the specificity of federation and its constraints. We also, propose a monitoring model that handles the complexity of managing a Multi-level monitoring and propose a solution to mitigate the cascading effect due to SLAs monitoring. Our monitoring scheme has the potential of efficiently reporting the source of performance violations, and its propagation to all dependent Cloud services. We evaluated our monitoring scheme using a series of experiments, and the results we have obtained are promising. They confirm that our scheme manages Multi-level SLAs in an efficient way, as it detects all violations, communicates these violations to concerned providers, and updates the SLAs accordingly.