云计算环境中面向时间约束的大规模并行业务流程的监控策略研究

Cloud computing can provide inexpensive, scalable, and high performance computing resources for time-constrained massive parallel business process systems, such as bank transferring and securities trading. However, It is difficult for cloud computing to complete these business processes timely to satisfy customers’ quality of service requirements, considering the dynamically changeable and unstable characters of cloud computing. Hence, it is necessary to design proper monitoring strategies in order to improve the throughput of business process systems. Besides, in scientific cloud workflows, these existing workflow activities’ response time-based sequence monitoring strategies are difficult to be deployed to massive parallel business processes directly. Based on the dynamic of cloud computing and concurrency of large scale business processes, this project proposes a novel throughput-based workflow temporal monitoring framework for a workflow lifecycle which consists of: 1) a throughput-based temporal modelling approach for parallel business processes; 2) a throughput-based temporal monitoring approach for parallel business processes; 3) a throughput-based temporal violation handling approach for parallel business processes. This strategy can timely detect intermediate temporal violations along the workflow execution and take effective control actions, so as to achieve better on-time completion rates for massive parallel business processes. Ultimately, the results of this project can help to significantly improve the efficiency in processing large numbers of customer requests.

云计算可以为银行转账、证券交易等大规模具有时间约束的业务系统提供廉价、可扩展和高性能的计算资源。然而，动态多变和不稳定性使得云计算中业务系统难以按时完成，不能满足用户服务质量要求。因此迫切需要设计合理的监控策略提高系统的吞吐量。而且，现有的科学云工作流中面向工作流活动响应时间的时序监控策略难以应用到大规模并行业务流程。因此，针对云计算的动态性和大规模业务流程的并发性，本项目以一个基于吞吐量的面向工作流生命周期的时序监控框架为手段，研究面向时间约束的大规模并行业务流程监控策略，包括：1）基于吞吐量的并行流程时序建模方法；2）基于吞吐量的并行流程时序监控方法；3）基于吞吐量的并行流程时序错误处理方法。该策略可以及时检测到执行过程中发生的时序错误并采取有效的处理措施，从而确保大规模并行业务流程的按时完成。本项目的研究成果将有效提高云环境中并行业务流程的按时完成率，从而更加高效的处理大规模用户请求

云计算可以为银行转账、证券交易等大规模具有时间约束的业务系统提供廉价、可扩展和高性能的计算资源。然而，动态多变和不稳定性使得云计算中业务系统难以按时完成，不能满足用户服务质量要求。因此迫切需要设计合理的监控策略提高系统的吞吐量。因此，针对云计算的动态性和大规模业务流程的并发性，本项目以基于吞吐量的面向工作流生命周期的时序监控框架为手段，主要从数据资源重新布局和任务重新调度两种策略角度，研究了大规模并行业务流程的时序一致性错误处理，获得了若干原创性成果。首先，研究了云计算环境中的科学工作流数据布局问题，建立了工作流层数据布局模型，提出了新的工作流层的智能数据布局算法。然后，研究了云计算环境中任务调度问题，建立了任务调度的能耗与时间验证模型，提出了解决任务调度问题的高效智能算法。最后，深入研究了云计算环境中资源分配问题，采用拍卖策略，建立了基于拍卖规则的云资源分配模型，提出了动态定价与迭代拍卖的资源分配策略。

内容获取失败，请点击重试