Narx-based Locally Distributed Web-servers Load Prediction

With the continuous development of network technology, the need to have a page view ofrnusers and the load on servers has grown exponentially, resulting in temporary loss of services.rnDistributed computing systems are becoming a widely used paradigm to provide highrnperformance and uninterrupted services to users. In such system, it is crucial to use effectivernresource management techniques to handle a large number of requests and providerndependable services with high quality constantly. Indeed, both service interruptions andrnresource waste can be reduced with the implementation of an eff ective prediction system.rnOne promising approach is realizing artificial neural network algorithm to predict resourcernusage series of server. Several research has been conducted using different combination ofrnload descriptor to predict resource usage series of server. Yet, queue time, is believed to be arngood load descriptor of a server, because it gives a good estimate of job response time. Itrnstrives to produce a global improvement in system performance. However, this loadrndescriptor, still not be seen in the study of resource usage prediction of server. Thus, in thisrnresearch, we have investigated nonlinear autoregressive network with exogenous inputsrn(NARX) neural network multi-step ahead predictability of web server load. Besides it, threerndifferent training algorithms: Lavenberg-Marquardt (LM), Bayesian Regularization (BR) andrnScaled Conjugate Gradient (SCG) forecasting accuracy were evaluated. We have collectedrnresource usage series of a week data from locally distributed web servers of ethio telecomrnbusiness support system. Two Cases (Case-1 and Case-2) of experiments were conducted tornevaluate the performance of the algorithms; using as input in the first Case CPU and memory,rnand in the second Case CPU, memory and queue time. MATLAB was employed to verify thernprediction accuracy of the algorithms. The results of the simulation show that for 12-steprnahead web server load prediction, LM learning algorithm, Case-2 approach has registered thernbest prediction accuracy with MAPE of 4.459%, followed by the BR learning algorithm withrnMAPE of 4.649%. SCG was the lowest performer with MAPE of 5.610%. Thus, accuracy inrnprediction is necessary since the more efficient resources can be managed in data centers.rnTherefore, having such a model, enhances the process of working towards reliable services.

Get Full Work

Report copyright infringement or plagiarism