This thesis investigates the application of autonomic management to adistributed storage system. Effects on performance and resource consumptionwere measured in experiments, which were carried out in a local area test-bed.The experiments were conducted with components of one specific distributedstorage system, but seek to be applicable to a wide range of such systems, inparticular those exposed to varying conditions. The perceived characteristicsof distributed storage systems depend on their configuration parameters and onvarious dynamic conditions. For a given set of conditions, one specificconfiguration may be better than another with respect to measures such asresource consumption and performance. Here, configuration parameter values wereset dynamically and the results compared with a static configuration. It washypothesised that under non-changing conditions this would allow the system toconverge on a configuration that was more suitable than any that could be set apriori. Furthermore, the system could react to a change in conditions byadopting a more appropriate configuration. Autonomic management was applied tothe peer-to-peer (P2P) and data retrieval components of ASA, a distributedstorage system. The effects were measured experimentally for various workloadand churn patterns. The management policies and mechanisms were implementedusing a generic autonomic management framework developed during this work. Theexperimental evaluations of autonomic management show promising results, andsuggest several future research topics. The findings of this thesis could beexploited in building other distributed storage systems that focus onharnessing storage on user workstations, since these are particularly likely tobe exposed to varying, unpredictable conditions.
展开▼
