【摘要】 分析表明,大数据任务的处理离不开可靠的设计框架,大数据的基本处理流程与传统数据处理流程并无太大差异。鉴于传统大数据应用框架在可用性和扩展性方面存在受限等不足,提出了一种具有良好适应性,可支持构建大数据应用的分层结构框架模型,该框架通过将实际应用涉及的大数据处理,抽象为若干可灵活组配和动态维护的数据处理流;每条数据处理流对应一个由若干称为介主体的计算节点构成的、支持并行执行的流式拓扑计算图.大量、位于不同层次、具有一定自适应能力的介主体,基于协约规则和所感知的环境知识协作,共同完成复杂的大数据处理任务。更多还原

【Abstract】 Dealing with tasks related with big data has important things to do with a reliable framework. The basic processes are similar to that of traditional data processing. Due to the shortcomings such as restricted reliability and expansibility of traditional big data related frameworks, amodel was assumed. A flexible and hierarchical framework model, which can be used to build complex big data applications, is developed. The actual complexed business big data processes are abstracted and modeled as several data processing flows that can be flexibly assembled and dynamically maintained. Each underlying data processing flow, which is called median-Agent, is corresponded to a topology graph that is comprised of a series of computing nodes, and can be executed parallelly. The application systems built based this framework, is typically composed of a large number of medianagents that are located on different layers, and have the adaptive ability to adjust and combine their behaviour to achieve complex data processing goals based on their environment knowledge perception and specific protocol rules.更多还原