arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2104.06274cs.DC
收藏

异构边缘云计算环境中数据共享科学工作流的最优数据放置

Optimal Data Placement for Data-Sharing Scientific Workflows in Heterogeneous Edge-Cloud Computing Environments

Xin Du, Songtao Tang, Zhihui Lu, Keke Gai, Jie Wu, Patrick C. K. Hung

摘要:

与传统的分布式计算或云计算环境相比,异构边缘-云计算范例可以为部署科学工作流提供更优化的方向。由于科学数据集的大小不一,且其中一些数据集必须保持私密性,因此寻找一种既能减少数据传输又能降低放置成本的数据放置策略仍然是一个难题。针对这一问题,结合边缘计算和云计算的优势,构建了一个数据放置模型,通过智能计算来平衡数据传输时间和数据放置成本。该模型解决的最困难的研究挑战是考虑在这种混合计算环境中的许多约束,其中包括在不同地理区域的单个和多个工作流之间的共享数据集。根据所建立的模型,提出了一种新的数据布局策略DE-DPSO-DPS,该策略采用差分进化离散粒子群优化算法(DE-DPSO-DPA)对这些科学数据集进行布局。该策略不仅考虑了边缘微处理器的数量和存储容量、不同微处理器之间的带宽以及私有数据集的比例等特性,还分析了算法在工作流执行过程中的性能。在模拟的异构边缘云计算环境中进行了综合实验,结果表明,与传统的数据共享科学工作流策略相比,该数据放置策略能够有效地减少数据传输时间和放置成本。

英文摘要:

The heterogeneous edge-cloud computing paradigm can provide a more optimal direction to deploy scientific workflows than traditional distributed computing or cloud computing environments. Due to the different sizes of scientific datasets and some of these datasets must keep private, it is still a difficult problem to finding an data placement strategy that can minimize data transmission as well as placement cost. To address this issue, this paper combines advantages of both edge and cloud computing to construct a data placement model, which can balance data transfer time and data placement cost using intelligent computation. The most difficult research challenge the model solved is to consider many constrain in this hybrid computing environments, which including shared datasets within individual and among multiple workflows across various geographical regions. According to the constructed model, the study propose a new data placement strategy named DE-DPSO-DPS, which using a discrete particle swarm optimization algorithm with differential evolution (DE-DPSO-DPA) to distribute these scientific datasets. The strategy also not only consider the characteristics such as the number and storage capacity of edge micro-datacenters, the bandwidth between different datacenters and the proportion of private datasets, but also analysis the performance of algorithm during the workflows execution. Comprehensive experiments are designed in simulated heterogeneous edge-cloud computing environments demonstrate that the data placement strategy can effectively reduce the data transmission time and placement cost as compared to traditional strategies for data-sharing scientific workflows.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑