基于数据集分割的云工作流模型库并行检索方法

Parallel Retrieval Approach of Cloud Workflow Model Repositories Based on Data Set Partitioning

在由多个行业云服务平台组成的集成服务平台中,随着行业云服务平台加盟数及各平台下租户数量的不断增多,其底层的云工作流模型库的规模也必将不断增大.当云工作流模型库的规模超大时,需要一种效率更高的并行检索方法去满足云工作流模型库高效检索的需求.鉴于此,采用均匀划分法或自动聚类法对大规模云工作流模型库进行合理的子集划分,并结合前期工作中已改进的基于图结构的流程检索算法,提出了基于数据集分割的大规模云工作流模型库并行检索方法.该方法主要包括4种流程并行检索算法:基于均匀划分模型集的静态并行检索算法、基于均匀划分模型集的动态并行检索算法、基于自动聚类模型集的静态并行检索算法和基于自动聚类模型集的动态并行检索算法.最后,在模拟生成的大规模流程集及真实的云工作流模型库中对这4种并行检索算法的检索效率进行了实验评估.

In the integrated service platform composed of multiple industry cloud service platforms, with the increasing of the number of cloud service platforms and theirs tenants, the scale of its underlying cloud workflow model repository will be increasing. When the scale of the cloud workflow model repository is super large, the existing retrieval methods of large-scale process model repositories still can＇t meet the needs of efficient retrieval of cloud workflow model repositories, therefore, it is necessary to study a more efficient parallel retrieval method. To address this issue, this paper adopts two data partitioning modes, equipartition and clustering based partitioning, to divide large-scale cloud workflow model repositories into small pieces. Combined with the improved process retrieval algorithm proposed in authors＇ previous work, a series of data partitioning based process parallel retrieval approaches are put forward to accelerate the large-scale process retrieval. These approaches mainly include four kinds of process retrieval algorithms from static/dynamic parallel retrieval algorithm based on uniform/automatic clustering partitioning model sets. Finally, based on the large-scale simulation process model library and the actual cloud workflow model repository, experiments are conducted to evaluate the efficiency of four parallel retrieval algorithms.