Novel management architecture for large-scale server cluster

As the size of the server cluster increases, management of clusters has been a daunting task. This paper proposed novel management architecture for large-scale server clusters, called cooperation-oriented hierarchical autonomic architecture (CHAA). CHAA emphasizes the autonomy of entities to implement intelligent management of server clusters. We detailed the architectures of CHAA and the intelligent agents. The resource tree based constraining mechanism has been presented to achieve conflict-free cooperation of multiple autonomous entities. Considering the uncertainty of agentspsila commitments, we introduced fuzzy belief and confidence degree into cooperation method of multi-agent. The measurements on confidence degree (cd) showed that agents have higher probabilities of gaining the best method or the best agent to cooperate with by introducing cd as a metric of cooperation.