面向复杂网络存储系统的元胞自动机动力学分析方法

Dynamics Analysis Method of Cellular Automata for Complex Networking Storage System

大规模网络存储系统中复杂的数据传输行为隐藏着一定的动力学规律性.针对基于对象的大规模网络存储系统,结合存储对象的智能性和主动性特征,分别在宏观与微观两个层次上提出了用于复杂网络存储动态行为规律分析的存储元胞自动机模型SNCA和OSDCA.在SNCA模型中,对网格拓扑结构的存储网络,结合存储对象的生命周期属性,可在宏观上分析网络存储系统的数据流动规律,确定存储网络拥塞程度,仿真结果揭示数据对象流动和存储网络中的相变具有全局相关性;在OSDCA模型中,综合热点数据的迁移和复制机制,在微观上分析I/O负载动态分布特性和存储热点迁移规律,仿真结果表明对象存储系统中的数据分布具有一定的自组织特性.

Some inherent dynamics rules are concealed in large-scale network storage system on account of the complexity of data transmission behaviors. This paper studies object-based storage system and proposes two storage cellular automata models called SNCA and OSDCA from macro and micro aspect respectively, and these models can be used to analyze behaviors and rules of complex and dynamic network storage by capturing the intelligent and initiative properties of storage object. In the model SNCA, the lifetime attribute of storage object is used to analyze the data flow rules in storage network to ascertain the congestion degree from macro aspect based on special lattice network topology architecture, and simulation results show that data object flow has global relativity with the phase transition of data flow. In the model OSDCA, data migration and replication mechanism are combined to analyze hotspot migration rule based on the load distribution condition among storage nodes, and simulation results show that data distribution in OBS system has characters of some self-organization.