摘要 |
The present invention relates to a method for processing jobs within a cluster architecture. The method comprises a step 901 of determining capacity at a node (101 see fig 1) before allocating a job for processing by the node (101) or adding the job to a cluster queue. A system for processing jobs within a cluster architecture is also disclosed. The nodes are connected via a communications system (104 see fig 1) such as a network. The invention is aimed at addressing situations such as when long processing times when the Internet is busy. Job allocation may include thresholds of CPU time, memory usage and total time, taking account of projected node loading over time. |