• Ujjal Aloke Sarkar
• Bhupender Kumar Singh

The horizontally extensible Internet service on a commercial computer cluster appears to be very suitable for automatic control. There is a target output (service-level agreement), an observation output (actual delay), and a gain controller (adjustment the number of servers). However, there are few data centers that are actually automated in this way in practice, due in part to well-founded skepticism about whether the simple models often used in the research literature can capture complex real-life workload/performance relationships and keep up with changing conditions that might invalidate the models. We agree these shortcomings can be solved by introducing modeling, control, and analysis techniques from statistics and machine learning. In particular, we applied a rich statistical model of application performance, a simulation-based approach to finding the optimal control strategy, and a change point to find abrupt changes in performance. Preliminary results of running a Web 2.0 benchmarking application driven by actual workload tracking in the Amazon EC2 cloud show that our approach can effectively control the number of servers, even in the face of performance anomalies is showing.

Web 2.0 benchmarking, Amazon EC2 cloud, statistical machine learning (SML)

"Automatic Control Realistic Through Statistical Machine Learning", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.6, Issue 6, page no.462-466, June 2019, Available :http://www.jetir.org/papers/JETIR1906Q65.pdf

"Automatic Control Realistic Through Statistical Machine Learning", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.6, Issue 6, page no. pp462-466, June 2019, Available at : http://www.jetir.org/papers/JETIR1906Q65.pdf