Providing Grid Schedulers with Passive Network Measurements

Abstract

Grids offer the potential to carry out difficult computing tasks and achieve superior aggregate performance through the utilization of remote resources and transparent collaboration between independent bodies (whether individuals or organizations). Grids are highly complex systems consisting of heterogeneous resources on disparate hosts interconnected across different Autonomous Systems and virtual organizations via a mixture of communication standards. Operating in such environments often proves erratic. Monitoring grid resources allows grid schedulers to adapt to changes in the status of remote resources and the network paths between them. Thus, monitoring the status of the different grid components is critical in order to ensure optimum performance. In this paper we introduce a distributed solution, GridMAP, to collect network and end-host resource measurements, analyze their performance and feed these statistics and predictions back to schedulers. At this stage, we present our implementation of a passive TCP-SYN technique to provide GridMAP with round trip time and throughput measurements and we evaluate our approach against ping and iperf.