| 摘要 |
<p><P>PROBLEM TO BE SOLVED: To provide an easy performance prediction method of a cluster system. <P>SOLUTION: In the method for predicting performance of the cluster system, a system configuration and operation state collection means 41 collects the number of unit of operation servers 311, a system operation rate, and measured values of system throughput to record them in a storage means 44 in order to predict performance of a cluster system 3 including a plurality of operation servers 311 which load-shares and performs processing of requests made in clients 1. A calculation means 42 reads the measured values to analyze them, and calculates a first amount of fluctuation in system throughput that is the amount of fluctuation in system throughput per a unit system operation rate in the same number of unit of the servers 311 and a second amount of fluctuation in system throughput that is the amount of fluctuation in system throughput per unit of server 311 in increasing or decreasing the number of unit of the servers 311 while maintaining the same system operation rate, and predicts the performance, etc., in varying the number of unit of the servers 311 of the cluster system by use of the calculated amounts of fluctuation in system throughput. <P>COPYRIGHT: (C)2007,JPO&INPIT</p> |
