Peer-to-peer systems have grown significantly in popularity over the last few years. An increasing number of research projects have been closely following this trend, looking at many of the paradigm's technical aspects. In the context of data-sharing services, efforts have focused on a variety of issues from object location and routing to fair sharing and peer lifespans. Overall, the majority of these projects have concentrated on either the whole P2P infrastructure or the client-side of peers. Little attention has been given to the peer's server-side, even when that side determines much of the everyday user's experience. In this paper, we make the case for looking at the server side of peers, focusing on the problem of scheduling with the intent of minimizing the average response time experienced by users. We start by characterizing server workload based on extensive trace collection and analysis. We then evaluate the performance and fairness of different scheduling policies through trace-driven simulations. Our results show that average response time can be dramatically reduced by more effectively scheduling the requests on the server-side of P2P systems.

Recently, size-based policies such as SRPT and FSP have been proposed for scheduling requests in web servers. SRPT and FSP are superior to policies that ignore request size, such as PS, in both efficiency and fairness given heavy-tailed service times. However, a central assumption that is usually made in implementing size-based policies in a web server is that the service time of a request is strongly correlated with the size of the file it serves. This paper shows how the performance of SRPT and FSP are affected by the degree of this correlation. We developed a simulator that supports both M/G/1/m and G/G/n/m queuing models. The simulator can be driven with trace data, which can be taken from the logs of modified Apache servers, or which can be produced by a workload generator we have developed that allows us to control the correlation. Using both trace data and generated data, we find that the degree of correlation has a dramatic effect on the performance of SRPT and FSP. In response, we propose and evaluate domain-based scheduling, a simple technique that better estimates connection times by making use of the source IP address of the request. Domain-based scheduling improves SRPT and FSP performance on web servers, particularly in regimes where correlation is low, thus making size-based policies such as these more broadly deployable.

Motivated by the optimality of Shortest Remaining Processing Time (SRPT) for mean response time, in recent years many computer systems have used the heuristic of “favoring small jobs” in order to dramatically reduce user response times. However, rarely do computer systems have knowledge of exact remaining sizes. In this paper, we introduce the class of ɛ-SMART policies, which formalizes the heuristic of “favoring small jobs” in a way that includes a wide range of policies that schedule using inexact job-size information. Examples of ɛ-SMART policies include (i) policies that use exact size information, e.g., SRPT and PSJF, (ii) policies that use job-size estimates, and (iii) policies that use a finite number of size-based priority levels. For many ɛ-SMART policies, e.g., SRPT with inexact jobsize information, there are no analytic results available in the literature. In this work, we prove four main results: we derive upper and lower bounds on the mean response time, the mean slowdown, the response-time tail, and the conditional response time of ɛ-SMART policies. In each case, the results explicitly characterize the tradeoff between the accuracy of the job-size information used to prioritize and the performance of the resulting policy. Thus, the results provide designers an understanding of how accurate job-size information must be in order to achieve desired performance guarantees.

Load balancing is a common approach to task assignment in distributed architectures such as web server farms, database systems, grid computing clusters, and others. In such designs there is a dispatcher that seeks to balance the assignment of service requests

The Shortest Remaining Processing Time (SRPT) scheduling policy was proven, in the 1960s, to yield the smallest mean response time, and recently it was proven its performance gain over Processor Sharing (PS) usually does not come at the expense of large jobs. However, despite the many advantages of SRPT scheduling, it is not widely applied. One important reason for the sporadic application of SRPT scheduling is that accurate job size information is often unavailable. Our previous work addressed the performance and fairness issues of SRPT scheduling when job size information is inaccurate. We found that SRPT (and FSP) scheduling outperforms PS as long as there exists a (rather small) amount of correlation between the estimated job size and the actual job size. In the work we summarize here, we have developed job size estimation techniques to support the application of SRPT to web server and Peer-to-Peer server side scheduling. We have evaluated our techniques with extensive simulation studies and real world implementation and measurement.

Recently, there have been a number of scheduling success stories in computer applications. Across a wide array of applications, the simple heuristic of “prioritizing small jobs ” has been used to reduce user response times with enormous success. As a result of the attention given to size based policies by computer systems

Abstract—Predicting grid performance is a complex task because heterogeneous resource nodes are involved in a distributed environment. Long execution workload on a grid is even harder to predict due to heavy load fluctuations. In this paper, we use Kalman filter to minimize the prediction errors. We apply Savitzky-Golay filter to train a sequence of confidence windows. The purpose is to smooth the prediction process from being disturbed by load fluctuations. We present a new adaptive hybrid method (AHModel) for load prediction guided by trained confidence windows. We test the effectiveness of this new prediction scheme with real-life workload traces on the AuverGrid and Grid5000 in France. Both theoretical and experimental results are reported in this paper. As the lookahead span increases from 10 to 50 steps (5 minutes per step), the AHModel predicts the grid workload with a mean-square error (MSE) of 0.04-0.73 percent, compared with 2.54-30.2 percent in using the static point value autoregression (AR) prediction method. The significant gain in prediction accuracy makes the new model very attractive to predict Grid performance. The model was proved especially effective to predict large workload that demands very long execution time, such as exceeding 4 hours on the Grid5000 over 5,000 processors. With minor changes of some system parameters, the AHModel can apply to other computational grids as well. At the end, we discuss extended research issues and tool development for Grid performance prediction.

We show how to significantly improve the mean response time seen by both uploaders and downloaders in peer-to-peer data-sharing systems. Our work is motivated by the observation that response times are largely determined by the performance of the peers serving the requested objects, that is, by the peers in their capacity as servers. With this in mind, we take a close look at this server side of peers, characterizing its workload by collecting and examining an extensive set of traces. Using trace-driven simulation, we demonstrate the promise and potential problems with scheduling policies based on shortest-remaining-processing-time (SRPT), the algorithm known to be optimal for minimizing mean response time. The key challenge to using SRPT in this context is determining request service times. In addressing this challenge, we introduce two new estimators that enable predictive SRPT scheduling policies that closely approach the performance of ideal SRPT. We evaluate our approach through extensive single-server and system-level simulation coupled with real Internet deployment and experimentation.