作者: Mauro Andreolini , Michele Colajanni , Marcello Pietri , Stefania Tosi
DOI: 10.1109/CNSM.2013.6727811
关键词: Adaptive algorithm 、 Distributed computing 、 Scalability 、 Cloud computing 、 Adaptive sampling 、 Computer science 、 Overhead (computing) 、 Real-time computing 、 Sampling (statistics) 、 System monitoring 、 Server
摘要: In large scale systems, real-time monitoring of hardware and software resources is a crucial means for any management purpose. architectures consisting thousands servers hundreds component resources, the amount data monitored at high sampling frequencies represents an overhead on system performance communication, while reducing may cause quality degradation. We present adaptive algorithm scalable that able to adapt frequency updating twofold goal: minimize computational communication costs, guarantee reduced samples do not affect accuracy information about resources. Experiments carried out heterogeneous traces referring synthetic real environments confirm proposed approach reduces utilization without penalizing with respect existing algorithms.