Finding the near workload type between TPC-C and TPC-W environments

초록

Database system can be performed more than a database application and shown peculiar workload characteristics as a database application. However, as a database system can be performed a lot of database applications, detecting workload characteristics might be more difficult. This paper proposes the method of finding near workload characteristics by mixed workloads. The workload find method is evaluated by the modified k-NN algorithm that is an extension of the existing the fuzzy k-NN algorithm. The modified k-NN algorithm measures how close workloads in TPC-C or TPC-W are to the mixed workloads between TPC-C and TPC-W. Furthermore, the result of our method is better than others for finding near workloads because its results are lower than others in the oscillation depending on the k parameter and the error rate. This study is able to contribute to finding similar information for providing autonomic database analysis and intelligent service in ubiquitous environments. © 2008 IEEE.