Transparent network services via a virtual traffic layer for virtual machines
Proceedings of the 16th international symposium on High performance distributed computing
TM-DG: a trust model based on computer users' daily behavior for desktop grid platform
Proceedings of the 2007 symposium on Component and framework technology in high-performance and scientific computing
Dynamic service selection in workflows using performance data
Scientific Programming - Dynamic Computational Workflows: Discovery, Optimization and Scheduling
Free network measurement for adaptive virtualized distributed computing
IPDPS'06 Proceedings of the 20th international conference on Parallel and distributed processing
Applying principles of active available bandwidth algorithms to passive TCP traces
PAM'05 Proceedings of the 6th international conference on Passive and Active Network Measurement
Hi-index | 0.00 |
Adaptive grid applications require up-to-date network resource measurements and predictions to help steer their adaptation to meet performance goals. To this end, we are interested in monitoring the available bandwidth of the underlying networks in the most accurate and least obtrusive way. Bandwidth is either measured by actively injecting data probes into the network or by passively monitoring existing traffic, but there is a definite trade-off between the active approach, which is invasive, and the passive approach, which is rendered ineffective during periods of network idleness. We are developing the Wren bandwidth monitoring tool, which uses packet traces of existing application traffic to measure available bandwidth. In this paper, we demonstrate that the principles supporting active bandwidth tools can be applied to passive traces of the LAN and WAN traffic generated by high-performance grid applications. We use our results to form a preliminary characterization of the application traffic required by available bandwidth techniques to produce effective measurements. Our results indicate that a low overhead, passive monitoring system supplemented with active measurements can be built to obtain a complete picture of the network's performance.