作者: Zach Hill , Jonathan Rowanhill , Anh Nguyen-Tuong , Glenn Wasson , John Knight
DOI: 10.1109/GRID.2007.4354131
关键词:
摘要: In order for Grids to become relied upon critical infrastructure and scientific computing, Grid-wide management must be automated so that it is possible in quickly comprehensively respond or anticipate specific environmental changes requirements. That is, due the disjoint administration of which results high communication requirements as well large numbers skilled administrators, manual reconfiguration portions a Grid such situations not viable solution. We have developed an architecture allows grid configuration response operator input sensors placed throughout Grid. Additionally, our resource owner specify who allowed alter his what types reconfigurations are allowed. This permits automation while retaining control individual resources. Our experiments show this can used reconfigure representative GT4-based development enforce maximum latency limits on Virtual Organization's jobs by dynamically controlling job priorities Grid-wide.