Measuring bottleneck link speed in packet-switched networks
Performance Evaluation
The macroscopic behavior of the TCP congestion avoidance algorithm
ACM SIGCOMM Computer Communication Review
Multivariate resource performance forecasting in the network weather service
Proceedings of the 2002 ACM/IEEE conference on Supercomputing
Combining active and passive network measurements to build scalable monitoring systems on the grid
ACM SIGMETRICS Performance Evaluation Review
Forecasting network performance to support dynamic scheduling using the network weather service
HPDC '97 Proceedings of the 6th IEEE International Symposium on High Performance Distributed Computing
Multigrain Parallelism for Eigenvalue Computations on Networks of Clusters
HPDC '02 Proceedings of the 11th IEEE International Symposium on High Performance Distributed Computing
Bandwidth Monitoring for Network-Aware Applications
HPDC '01 Proceedings of the 10th IEEE International Symposium on High Performance Distributed Computing
Network Characterization Service (NCS)
HPDC '01 Proceedings of the 10th IEEE International Symposium on High Performance Distributed Computing
Packet-dispersion techniques and a capacity-estimation methodology
IEEE/ACM Transactions on Networking (TON)
Nettimer: a tool for measuring bottleneck link, bandwidth
USITS'01 Proceedings of the 3rd conference on USENIX Symposium on Internet Technologies and Systems - Volume 3
Measuring and characterizing system behavior using kernel-level event logging
ATEC '00 Proceedings of the annual conference on USENIX Annual Technical Conference
The PingER project: active Internet performance monitoring for the HENP community
IEEE Communications Magazine
IMPuLSE: integrated monitoring and profiling for large-scale environments
LCR '04 Proceedings of the 7th workshop on Workshop on languages, compilers, and run-time support for scalable systems
Task-execution scheduling schemes for network measurement and monitoring
Computer Communications
Hi-index | 0.00 |
Distributed applications require timely network measurementsso that they can adapt to changing network conditionsand make efficient use of grid resources. One of thekey issues in obtaining network measurements is the intrusivenessof the measurements themselves-how much networkperformance is "wasted" taking the measurements?Our goal is to combine active and passive monitoring techniquesto reduce the need for intrusive measurements withoutsacrificing the accuracy of the measurements. We aredeveloping a bandwidth monitoring tool as part of the Wrennetwork measurement system that will reduce the burden onthe network by passively obtaining measurements from existingapplication traffic whenever possible, instead of activelyprobing the network. By using passive measurementswhen an application is running and active measurementswhen none are running, we can offer accurate, timely availablebandwidth measurements while limiting the invasivenessof active probes. We have completed a prototype ofthe Wren bandwidth monitoring tool and present our preliminaryanalysis of its performance in this paper. We provideresults from passive implementations of several availablebandwidth techniques and demonstrate the close quantitativerelationship between the results of both active andpassive techniques. We have tested our implementation ina cluster, across a campus, and across the Internet usingbulk data transfers as well as an adaptive eigenvalue application.Our results with this diverse set of environmentsand traffic types show promise toward implementing thesetechniques as measurement services in production environments.