Extensibility Challenges of Scientific Workflow Management Systems.

HCI (6)(2023)

引用 0|浏览8
暂无评分
摘要
Researchers compose scientific workflows for complex scientific experiments and simulations by connecting tools and data in a pipeline. The usability of a scientific workflow management system (SWfMS) largely depends on the availability of necessary tools in the system and the simplicity of their usage. Scientific experiments are incredibly diverse and need a wide variety of tools. Due to an overwhelming number of available tools in the public domain, a SWfMS cannot preinstall all tools required for multifarious experiments. Hence an extensibility mechanism to integrate external tools is greatly important for the flexibility of SWfMS. Tools are independently developed by different development teams using their favorite or suitable programming languages and may run on different operating environments. The tool integration is challenging due to the myriad development languages used for tools and potentially varying operating environments of SWfMS and tools. The software tools may not be robust enough for workflow integration. The state-of-the-art SWfMSs such as Galaxy and KNIME are web-based and can simultaneously serve hundreds of users. The end-users may want to quickly integrate their code in a SWfMS as a tool and use it in a workflow model. But many tools require a system configuration change, which end-users are not authorized to do. The integrated tool must also fit the workflow pipeline with input and output datasets. End-users need an efficient user interface for tool integration by themselves. We created 50 workflows in image processing, bioinformatics, and software analytics domains using VizSciFlow SWfMS. We gathered the challenges we encountered while extending it by integrating tools for these workflows using its extensibility interface. In this paper, we describe the challenges and propose solutions with the help of two case studies we conducted by developing two real-world workflow products - CoGe’s SynMap workflow in the Bioinformatics domain and source code clone detection and validation in the Software Analytics domain.
更多
查看译文
关键词
workflow,management,scientific,systems
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要