Distributed analytics technique for monitoring data
Radu Alexandru Badea, Serban Georgica Obreja · 2016
Monitoring the activity of equipment in datacenters is often a key element in designing large scale network and IT infrastructure. The information collected by direct parameter observations (such as total network traffic, CPU usage, etc.) represents a first step in the overall system performance optimization. A second one is denoted by data analytics and decision making strategies based on the values obtained in the (parameter) acquisition phase. The computing effort required to process the stored data can be sometimes important, setting a necessary but costly burden upon the existing datacenter analytics infrastructure. To address this issue, the present paper proposes a decentralized and cooperating architecture where end user machines will execute at least partially some of the data analytics code and return subsequently the results to managing server units in the datacenter.