分布式多模型在线预测(DMOP)机制的设计与实现

The Design and Realization of the Mechanism of Distributed Multi-Model Online Prediction(DMOP)

本文通过对海量信息处理集群系统的负载特性进行收集和分析,提出了一种分布式多模型在线预测(DMOP)机制,该机制包含四个关键模块:向后滑动聚合模块、负载图样评估模块、算法库模块和预测误差评估模块。本文通过试验对模块实现进行了测试,证明了模块设计的有效性和实用性。在DMOP机制算法库模块架构下,本文提出了一种改进的AR(p)(自回归模型)预测算法,并通过了测试。

This paper reviews the current research on host load prediction systematically and analyzes the advantages and disadvantages of various technologies in the first place. Then presents a mechanism called I）MOP （distributed multi-model online prediction） ; We confirms the effectiveness and feasibility of the realization of the mechanism via a large number of tests. Moreover, this paper proposes a improved AR（p） model prediction algorithm in the DMOP mechanism algorithm library module structure and tested its realization.