In the ever-growing grid infrastructures picking the right resources is not an easy task. There are situations were finding enough idle CPU's that satisfy a set of minimum nominal hardware requirements is simply not enough. In many such occasions, users or VO managers need to be involved in the resource selection process, either by excluding resources or by indicating preference, more so when resources are not for free. To support users in this task, we propose an interactive and user-driven approach to a performance-based ranking of grid resources. Using the GridBench framework and the provided suite of micro-benchmarks, the user is able to interactively explore and rank what is currently available on the grid in order to pinpoint the right resources. Through this framework, the user is able to interactively compose ranked lists of grid resources using custom ranking functions that are based on low-level measurements
[1]
Marios D. Dikaiakos,et al.
Grid benchmarking: vision, challenges, and current status
,
2007,
Concurr. Comput. Pract. Exp..
[2]
Marios D. Dikaiakos,et al.
GridBench: A Workbench for Grid Benchmarking
,
2005,
EGC.
[3]
Marios D. Dikaiakos,et al.
GridBench: a tool for benchmarking grids
,
2003,
Proceedings. First Latin American Web Congress.
[4]
Marios D. Dikaiakos.
Grid benchmarking: vision, challenges, and current status: Research Articles
,
2007
.
[5]
Marios D. Dikaiakos,et al.
Characterization of Computational Grid Resources Using Low-Level Benchmarks
,
2006,
2006 Second IEEE International Conference on e-Science and Grid Computing (e-Science'06).