A Goal-Oriented Approach for Adaptive SLA Monitoring : a Cloud Provider Case Study

Abstract

We argue in this paper that autonomic systems need to make their integrated monitoring adaptive in order to improve their “comprehensive” Quality of Service (QoS). We propose to design this adaptation based on high level objectives (called goals) related to the management of both the “functional system QoS” and the “monitoring system QoS”. Starting from some previous works suggesting a model-driven adaptable monitoring framework composed of 3 layers (configurability, adaptability, governability), we introduce a methodology to identify the functional and monitoring high level goals (according to the agreed Service Level Agreement - SLA) in order to drive models' instantiation. This proposal is first applied to a cloud provider case study for which two high level goals are developed (respect metrics freshness and minimize monitoring cost), and then simulated to show how the quality of management decisions, as well as intelligent monitoring of dynamic SLA, could be improved.