Scientific workflow management systems and workflow patterns
Abstract
Scientific workflow management systems are primarily used by scientists to design workflows that underlie many complex e-science applications. As Gil, et al. (2007) observed, the applications of these workflows and the distributed infrastructure over which these workflows operate are always in a state of flux. Hence the workflows need to be dynamic in nature wherein the decision about which steps to take next is based on the previous results. The current scientific workflow management systems do not provide enough support for dynamic workflows (Gil, et al., 2007). In recent years a number of scientific workflow management systems have surfaced with each having their own ways of doing the same set of things. Due to this there is a lack of standardization among them (Zhao, Raicu, & Foster, 2008). The current workflow development is mostly customized and there is hardly any reuse. Such reuse would lead to rapid application development and also increase the correctness of the workflows developed (Xiang and Madey, 2007). In software development, design patterns (Gamma, Helm, Johnson, & Vlissides, 1995) have served as a standard specification at the design level and have lead to design reuse. Similarly there is a catalogue of commonly occurring control flow constructs in workflow management systems called workflow patterns (Aalst, Hofstede, Kiepuszewski, & Barros, 2003). Some of these patterns provide a certain degree of dynamicity as well. This project proposes to implement these constructs in scientific workflow management system with the aim that the introduction of these patterns will provide a standard specification at the design level that can be reused and thus lead to an efficient and robust development process.
Degree
M.S.
Advisors
Springer, Purdue University.
Subject Area
Computer science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.