A highly accurate monitoring solution for active network measurement is provided without the need for GPS, based on an alternative software clock for PC's running Unix. With respect to clock rate, it's perform...
详细信息
ISBN:
(纸本)9781581135312
A highly accurate monitoring solution for active network measurement is provided without the need for GPS, based on an alternative software clock for PC's running Unix. With respect to clock rate, it's performance exceeds common GPS and NTP synchronized software clock accuracy. It is based on the TSC register counting CPU cycles and offers a resolution of around 1ns, a rate stability of 0.1PPM equal to that of the underlying hardware, and a processing overhead well under 1µs per timestamp. It is scalable and can be run in parallel with the usual clock. It is argued that accurate rate, and not synchronised offset, is the key requirement of a clock for network measurement. The clock requires an accurate estimation of the CPU cycle period. Two calibration methods which do not require a reference clock at the calibration point are given. To the TSC clock we add timestamping optimisations to create two high accuracy monitors, one based on Linux and the other on Real-Time Linux. The TSC-RT-Linux monitor has offset fluctuations of the order of 1µs. The clock is ideally suited for high precision active measurement.
NAMD is a fully featured, production molecular dynamics program for high performance simulation of large biomolecular systems. We have previously, at SC2000, presented scaling results for simulations with cutoff elect...
详细信息
The proceedings contains 46 papers from the conference on Performance Evaluation review. The topics discussed included: network routing and forwarding;system performance;mobile and wireless networks;network protocols ...
详细信息
The proceedings contains 46 papers from the conference on Performance Evaluation review. The topics discussed included: network routing and forwarding;system performance;mobile and wireless networks;network protocols and modeling;memory systems;transmision control protocol;web performance analysis techniques;automation and modeling of wireless networks.
The measurement, analysis and implementation of a working video server testbed implementing both patching and periodic broadcast algorithms was reported. The testbed consisted of both Linux and Windows-based clients a...
详细信息
The measurement, analysis and implementation of a working video server testbed implementing both patching and periodic broadcast algorithms was reported. The testbed consisted of both Linux and Windows-based clients and a Linux-based, application level video server. The underlying system issues were investigated that arose when putting idealized algorithms into practice.
We propose a user model that explains the shape of the distribution of file sizes in local file systems and in the World Wide Web. We examine evidence from 562 file systems, 38 web clients and 6 web servers, and find ...
详细信息
We propose a user model that explains the shape of the distribution of file sizes in local file systems and in the World Wide Web. We examine evidence from 562 file systems, 38 web clients and 6 web servers, and find that the model is a good description of these systems. These results cast doubt on the widespread view that the distribution of file sizes is long-tailed and that long-tailed distributions are the cause of self-similarity in the Internet.
A framework to unify and generalize the max-min fairness, variance related measures and average utility measures to characterize the protocol fairness using approximate majorization was presented. The theory was illus...
详细信息
A framework to unify and generalize the max-min fairness, variance related measures and average utility measures to characterize the protocol fairness using approximate majorization was presented. The theory was illustrated in the context of networking protocols with a simple case study of the fairness of network protocols. The theory aided in the evolution of the fairness of existing and helped protocol designing.
The performance of periodic batch rekeying in a group communication system was analyzed based on the use of key trees. A reliable rekey transport protocol was designed and the server and client bandwidth requirements ...
详细信息
The performance of periodic batch rekeying in a group communication system was analyzed based on the use of key trees. A reliable rekey transport protocol was designed and the server and client bandwidth requirements for the protocol were evaluated. The system improved the synchronization problem and improved rekey encoding efficiency by rekeying periodically for a batch of join/leave requests.
The running time advisor (RTA), which can predict the running time of a compute-bound task on a shared host running a variant of the Unix operating system, was described. These predictions are valuable for scheduling ...
详细信息
The running time advisor (RTA), which can predict the running time of a compute-bound task on a shared host running a variant of the Unix operating system, was described. These predictions are valuable for scheduling the soft real-time tasks of distributed interactive applications such as scientific visualizations. The RTA predicts running time as a confidence interval computed to the application's requested confidence level. Confidence levels provide a simple abstraction to the application and even provide information to enable valid statistical reasoning in the scheduling process.
The log data analyses for two real-time streaming media servers at public universities in the USA were discussed. The three goals to analyze the server logs are: providing data for generating synthetic workloads;cachi...
详细信息
The log data analyses for two real-time streaming media servers at public universities in the USA were discussed. The three goals to analyze the server logs are: providing data for generating synthetic workloads;caching strategies;and quantifying the effectiveness of multicast streaming methods. The time between interactive requests in the eTeach server follows heavy-tailed distribution, while distribution of file popularity can be modeled by the concatenation of two Zipf-like distribution. High client interactivity was observed in eTeach, which can be reduced by 40-60% using multicast delivery.
Two performance tools supporting application performance tuning are described. In application performance tuning, many types of information with source code are correlated to locate and analyze performance problems bo...
详细信息
Two performance tools supporting application performance tuning are described. In application performance tuning, many types of information with source code are correlated to locate and analyze performance problems bottlenecks. The two tools, MHSim and HPCView produce their output as HTML databases that can be analyzed using a commodity browser. MHSim investigates problems with poor memory hierarchy utilization in Fortran programs. HPCView produces a navigable HTML document which combines performnace metrics from different sources. New metrices are then synthesized and correlated with program source code.
暂无评论