Daniel W. Yoas
ABSTRACT
Short-term prediction has been established in computing as a mechanism for improving services. Long-term prediction has not been pursued because attempts to use multiple steps to extend short-term predictions have been shown to become less accurate the further into the future the prediction is extended. In each case, the researchers used fine grained sampling for the analysis. This study used course sampling of ten-second intervals and then aggregated them into periods of minutes, fifteen-minutes, and hours. Each of the aggregates was used to calculate the predictions for Hourly, Daily, and Weekly cycles, determine the error rate of the prediction, and establish a confidence interval of 80%. The results then were evaluated to identify the effectiveness of long term prediction and the best cycle to predict the resource utilization most accurately.
- Abusina, Z. U. M., S. M. S. Zabir, A. Ashir, D. Chakraborty, T. Suganuma, and N. Shiratori (2005) An engineering approach to dynamic prediction of network performance from application logs. Int. J. Netw. Manag. 15, 151--162 DOI: 10.1002/nem.554. Google Scholar
Digital Library
- Andreolini, M. and S. Casolari (2006) Load prediction models in web-based systems. Proceedings of the 1st international conference on Performance evaluation methodolgies and tools, 27 DOI: 10.1145/1190095.1190129. Google ScholarDigital Library
- Andreolini, M., S. Casolari, and M. Colajanni (2008) Models and framework for supporting runtime decisions in Web-based systems. ACM Trans. Web 2, 1--43 DOI: 10.1145/1377488.1377491. Google ScholarDigital Library
- Barford, P. and M. Crovella (1998) Generating representative Web workloads for network and server performance evaluation. SIGMETRICS Perform. Eval. Rev. 26, 151--160 DOI: 10.1145/277858.277897. Google ScholarDigital Library
- Dinda, P. A. (2002) A Prediction-Based Real-Time Scheduling Advisor. International Parallel and Distributed Processing Symposium (IPDPS'02) 1, 0010b-0010b DOI: 10.1109/IPDPS.2002.1015480. Google Scholar
- Krithikaivasan, B., Y. Zeng, K. Deka, and D. Medhi (2007). ARCH-based traffic forecasting and dynamic bandwidth provisioning for periodically measured nonstationary traffic. IEEE/ACM Trans. Netw. 15, 683--696 DOI: 10.1109/tnet.2007.893217. Google ScholarDigital Library
- Sharifian, S., S. A. Motamedi, and M. K. Akbari (2010). An approximation-based load-balancing algorithm with admission control for cluster web servers with dynamic workloads. J. Supercomput. 53, 440--463 DOI: 10.1007/s11227-009-0303-8. Google ScholarDigital Library
- Wolski, R. (2003) Experiences with predicting resource performance on-line in computational grid settings. SIGMETRICS Perform. Eval. Rev. 30, 41--49 DOI: 10.1145/773056.773064. Google ScholarDigital Library
Index Terms
- Resource utilization prediction: long term network web service traffic
Recommendations
Resource utilization prediction: a proposal for information technology research
RIIT '12: Proceedings of the 1st Annual conference on Research in information technologyResearch into predicting long-term resource needs has been faced with a very difficult problem of extending the accuracy period beyond the immediate future. Business forecasting has overcome this limitation by successfully incorporating the concept of ...
Using time-series analysis to provide long-term CPU utilization prediction
RIIT '14: Proceedings of the 3rd annual conference on Research in information technologyTime-series analysis has been a recognized method of prediction for years. In computing, prediction has revolved around scheduling problems and is used in computing time to manage resources in the short-term. Long-term prediction using these methods has ...
CPU utilization prediction methods overview
CEE-SECR '15: Proceedings of the 11th Central & Eastern European Software Engineering Conference in RussiaCPU utilization prediction is an important component of the resource allocation and job scheduling problems in virtualized servers and clouds. Utilization prediction is used by a resource manager to balance workload among physical nodes thus increasing ...
Comments