A Terminology for Scientific Workflow Systems Journal: arXiv
Frédéric Suter, Tainã Coleman, İlkay Altintaş, Rosa M. Badia, Bartosz Balis, Kyle Chard, Iacopo Colonnelli, Ewa Deelman, Paolo Di Tommaso, Thomas Fahringer, Carole Goble, Shantenu Jha, Daniel S. Katz, Johannes Köster, Ulf Leser, Kshitij Mehta, Hilary Oliver, J.-Luc Peterson, Giovanni Pizzi, Loïc Pot · 2025
Read A Terminology for Scientific Workflow Systems Journal: arXiv by Frédéric Suter, Tainã Coleman, İlkay Altintaş, Rosa M. Badia, Bartosz Balis, Kyle Chard, Iacopo Colonnelli, Ewa Deelman, Paolo Di Tommaso, Thomas Fahringer, Carole Goble, Shantenu Jha, Daniel S. Katz, Johannes Köster, Ulf Leser, Kshitij Mehta, Hilary Oliver, J.-Luc Peterson, Giovanni Pizzi, Loïc Pot (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-07838.txt.
Introduction
Frédéric Suter and co-authors build a shared language for scientific workflow systems in a 2025 paper. A workflow is a chain of linked computing tasks. Hundreds of systems now exist to run such chains. With no shared terms, researchers struggle to compare systems or pick the right one for a job.
A group of system builders and users agreed on a terminology with five axes. The axes cover what the workflow looks like, how it is composed, and how it is run. They also cover how data moves and what records are kept. Each axis holds concepts that capture the traits that matter most. The authors then classify 23 existing systems against the axes. The result is a map meant to guide both users choosing a system and developers describing one.
Catalog record
| CREATOR | Frédéric Suter, Tainã Coleman, İlkay Altintaş, Rosa M. Badia, Bartosz Balis, Kyle Chard, Iacopo Colonnelli, Ewa Deelman, Paolo Di Tommaso, Thomas Fahringer, Carole Goble, Shantenu Jha, Daniel S. Katz, Johannes Köster, Ulf Leser, Kshitij Mehta, Hilary Oliver, J.-Luc Peterson, Giovanni Pizzi, Loïc Pot |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 51 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-07838
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A Terminology for Scientific Workflow Systems Journal: arXiv Frédéric Suter, Tainã Coleman, İlkay Altintaş, Rosa M. Badia, Bartosz Balis, Kyle Chard, Iacopo Colonnelli, Ewa Deelman, Paolo Di Tommaso, Thomas Fahringer, Carole Goble, Shantenu Jha, Daniel S. Katz, Johannes Köster, Ulf Leser, Kshitij Mehta, Hilary Oliver, J.-Luc Peterson, Giovanni Pizzi, Loïc Pottier, Raül Sirvent, Eric Suchyta, Douglas Thain, Sean R. Wilkinson, Justin M. Wozniak, Rafael Ferreira da Silva Abstract The term “scientific workflow” has evolved over the last two decades to encompass a broad range of compositions of interdependent compute tasks and data movements. It has also become an umbrella term for processing in modern scientific applications. Today, many scientific applications can be considered as workflows made of multiple dependent steps, and hundreds of workflow systems have been developed to manage and run these scientific workflows. However, no turnkey solution has emerged from the field to address the diversity of scientific processes and the infrastructure on which they are supposed to be implemented. Instead, new research problems requiring the execution of scientific workflows with some novel feature often lead to the development of an entirely new workflow system. A direct consequence of this situation is that many existing workflow management systems (WMSs) share some salient features, offer similar functionalities, and can manage the same categories of workflows but at the same time also have some distinct capabilities that can be important for specific applications. This situation makes researchers who develop workflows face the complex question of selecting a WMS. This selection can be driven by technical considerations, to find the system that is the most appropriate for their application and for the computing and storage resources available to them, or other factors such as reputation, adoption, strong community support, or long-term sustainability. To address this problem, a group of WMS developers and practitioners joined their efforts to produce a community-based terminology of WMSs. This paper summarizes their findings and introduces this new terminology to characterize WMSs. This terminology is composed of fives axes: workflow structure and characteristics, composition, orchestration, data management, and metadata capture. Each axis comprises several concepts that capture the prominent features of WMSs. Based on this terminology, this paper also presents a classification of 23 existing WMSs according to the proposed axes and terms. 1 Introduction The concept of workflows , i.e., the execution of orchestrated and repeatable patterns of activity, dates back to the early 1900s when the engineering and manufacturing community introduced one of the earliest examples of procedural workflow: the Ford assembly line adopted by automobile manufacturers to this date. Workflows has been used to model, analyze, and improve business processes, using tools such as flow charts, functional flow block diagrams, or control flow diagrams [ 1 ] . The database community has also used workflows to address the challenges of managing large datasets [ 2 ] . The capacity to describe and orchestrate such complex applications popularized workflows across multiple scientific domains. The term scientific workflow itself was introduced in 1996 [ 3 , 4 ] to differentiate this specific type of workflow from the business and automation pipelines that inspired them. As scientific workflow may designate processes that go beyond science to cover more broadly defined research activities, we opted for the use of the term workflow in the remainder of this article with the following all-encompassing definition: Definition. A workflow is a structured sequence of computational tasks or activities that achieve a research or analytical objective. Workflows define the flow of work, including the order of steps, the data and control dependencies between them, and the rules governing their execution. Modern workflows extend beyond traditional directed acyclic graphs to encompass dynamic, adaptive, and interactive processes that may include cycles, branches, and human-in-the-loop components. They span diverse domains, including scientific research, engineering, humanities, and business, and bridge heterogeneous computing environments from edge devices to high-performance computing (HPC) facilities and cloud infrastructure. Over the past decades, workflows have become the predominant format for describing complex, multi-step, multi-domain scientific applications [ 5 ] . To manage the composition, planning, orchestration, and automation of the efficient execution of such workflows on powerful and often distributed compute infrastructures, a wide variety of workflow management systems (WMSs) have been proposed [ 6 ] . However, domain researchers who develop workflows and want to rely on a WMS to execute them often face the complex question of selecting a particular WMS. This selection can be driven by technical considerations, such as finding the most appropriate system for their application and for the computing and storage resources available to them, or factors such as reputation, adoption, community support, or long-term sustainability. The main reasons for this challenge are that no single ideal turnkey solution has emerged from the field to address the diversity of scientific processes and the heterogeneity of possible execution environments (both in terms of hardware and software). Instead, new research problems or new computer technologies related to the execution of workflows often lead to the development of an entirely new workflow management system. Figure 1: Five axes that categorize workflows and workflow management systems, with each axis further delineated into corresponding terms and sub-terms to provide a structured and detailed terminology. A direct consequence of this situation is that many existing WMSs share some salient features, offer similar functionalities, and can manage the same categories of workflows, but often also have distinct features tailored for specific types of problems. This has been highlighted by different efforts to create taxonomies and characterizations of workflows and WMSs [ 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 ] . These efforts can help to provide workflow developers with some guidance when trying to select the appropriate tool to develop and execute their workflows, but they are also notoriously incomplete and quickly outdated in a fast-moving field. Consequently, decisions for specific systems are very often based on social aspects as much or more than on technical ones (e.g., previous experience in the community, word-of-mouth, comments in web forums, personal evaluation of a few known systems). Inspired by the work of the in situ processing community for data visualization and analysis systems [ 16 ] , we propose in this article to go beyond a traditional taxonomy and develop a consistent terminology to describe WMSs. . To this end, we gathered a group of workflow system developers and workflow practitioners, all members of the Workflows Community Initiative (WCI) [ 17 ] , and followed a process similar to that in [ 16 ] to create a strong terminology for WMSs. This paper synthesizes the discussions initiated during the different editions of the Workflows Community Summit [ 18 , 19 , 20 , 21 , 22 , 23 ] , which led to the writing of this paper. The main contribution of this paper is the identification of five axes to characterize WMSs (Figure 1 ). Each axis comprises a series of concepts that capture the most salient features of WMSs. Based on the proposed terminology, our group analyzed 23 actively developed WMSs that are part of the WCI to determine which combination of terms can define each of them. The remainder of this