基于随机Petri网的大数据计算模式的模型与评价

随着云计算、物联网、社交网络等技术的迅猛发展，数据的快速增长成为许多行业共同面临的严峻挑战，大数据得到越来越广泛的关注。海量数据的分析与处理，是大数据的一个重要方面。大数据计算模式，即大数据分析与处理的抽象模型及方法论，是大数据计算方法和系统的设计与优化中的核心问题。本项目从大数据计算的特性出发，结合大数据计算系统等应用实例，从操作模式、服务模式和管控模式三个方面，研究大数据计算模式。基于随机Petri网等模型工具，给出普适的大数据计算模式的抽象模型。结合大数据计算中用户的需求，在抽象模型基础上，建立大数据计算的性能和可信赖性不同维度指标的评价模型，研究相关的模型分析求解技术，给出形式化的量化分析。本项目的研究，将为大数据计算的流程设计、机制支撑与系统实现提供理论基础，同时也将丰富和发展基于随机Petri网的性能评价理论。

With the development of Cloud Computing, Internet of Things and social networks, the volume of data is growing rapidly. Big Data analytics becomes a critical problem in both industry and academia. The computational paradigm, which is the abstract modeling and methology of Big Data computing, is an important concern. In this project, based on the characteristics of Big Data computing and its applications, we study the computational paradigms for Big Data from three aspects, which are processing paradigm, service paradigm and control paradigm. We model the paradigms using stochastic Petri nets, and build performance and dependability models for the evaluation of Big Data computing. Furthermore, we study the model analytics for Petri nets, and give quantitative analysis of the evaluation metrics. We believe our project can provide theoretical and technical supports for the design and development of Big Data computing systems, and bring new theories and techniques to performance evaluation based on stochastic Petri nets.