In today's service market the provisioning of high quality services has become a critical issue for providers as the business success of their customers is often based on the well functioning of these subscribed services. Besides a proper service configuration an efficient fault management needs to be in place. Requirements for such a fault management are the timely resolution of problems affecting the service quality and a reasonable balance between the fault management effort and the costs saved by preventing service level agreement violations. In order to fulfill these requirements we propose the adaptation of event correlation techniques which have already proven to be useful in the area of network and systems management Our hybrid architecture consists of a rule-based reasoning module, whose rules are derived from a modeling of services and underlying infrastructure, and a case-based reasoning module. Due to the complexity of today's service provisioning we use the latter one to collect cases that cannot be covered by the rules so far. The experience gained from the cases is used to improve the modeling and therefore to improve the rules. We use a service provisioning scenario at a large IT service provider to show the applicability of our approach.
展开▼
