Affiliations: System Engineering Section, Policy and Management, Faculty of Technology, Delft University of Technology, The Netherlands
Correspondence:
[*]
Corresponding author: Kassidy P. Clark, System Engineering Section, Policy and Management, Faculty of Technology, Delft University of Technology, Jaffalaan 5, 2628BX Delft, The Netherlands. Tel.: +31 15 278 73 59; E-mail: k.p.clark@tudelft.nl
Note: [1] This paper is an extension and revision of a paper previously presented at the 2011 International Conference on Adaptive and Intelligent Systems [1].
Abstract: Cloud Service Providers offer access to hardware and software resources across the Internet. Cloud Service Providers and their clients negotiate Service Level Agreements (SLA) that specify terms and conditions of services. To ensure that an SLA is being honored by all parties the service is monitored. This paper proposes a monitoring approach that automatically adapts the monitoring process in realtime, based on user-specified policies. The monitor can both adapt the frequency of measurements and monitoring mode: active and passive. Active mode uses an external, trusted monitoring service. Passive mode minimizes interaction with an external monitoring service, relying instead on a mutual commit protocol to ensure SLA compliance. Monitor adaptation is determined by levels of perceived risk, defined by a user's policy. The self-adaptive SLA monitoring framework is realized using agent technology and is implemented and tested in the AgentScape middleware.
