| 摘要 |
PROBLEM TO BE SOLVED: To provide a cluster system capable of obtaining calculation processing capability corresponding with a scale even if a configuration is an inexpensive configuration. SOLUTION: In this cluster system 1 wherein a large number of job execution nodes 200 are connected to each other by a high-speed network 50, when a control server 100 allocates a job to each the job execution node 200, the control server 100 selects a 'node a1' and a 'node a2' similar to the 'node a1', makes data by the job executed by a node of the 'node a1' be stored in a node of the 'node a2' and a node itself of the 'node a1', and makes the data redundant, so that the data can be made to be redundant between the nodes while suppressing a load on the whole cluster to the minimum, and the calculation processing capability corresponding with the scale of the cluster can be obtained even if the configuration is the low-cost configuration. COPYRIGHT: (C)2008,JPO&INPIT
|
