Mirko Sonntag

Conventional Workflow Technology for Scientific Simulation

Workflow technology is established in the business domain for several years. This fact suggests the need for detailed investigations in the qualification of conventional workflow technology for the evolving application domain of e-Science. This chapter discusses the requirements on scientific workflows, the state of the art of scientific workflow management systems as well as the ability of conventional workflow technology to fulfill requirements of scientists and scientific applications. It becomes clear that the features of conventional workflows can be advantageous for scientists but also that thorough enhancements are needed. We therefore propose a conceptual architecture for scientific workflow management systems based on the business workflow technology as well as extensions of existing workflow concepts in order to improve the ability of established workflow technology to be applied in the scientific domain with focus on scientific simulations.

Next Generation Interactive Scientific Experimenting based on the Workflow Technology

In this paper we explore to what extent the conventional workflow technology and service-oriented architecture (SOA) principles can be applied to support scientist in their experiments. Based on the requirements imposed on systems for scientific computing, e-Science and simulations, and an extended workflow life cycle we introduce the architecture of an interactive system that reuses the conventional workflow technology. We advocate the realization of this workflow system with advanced adaptation and monitoring features because we identified that modeling of scientific applications and simulations can only be done the “scientists’ way” if the traditional workflow modeling as well as design and run time adaptation are combined in a user-friendly solution.

Using services and service compositions to enable the distributed execution of legacy simulation applications

In the field of natural and engineering science, computer simulations play an increasingly important role to explain or predict phenomena of the real world. Although the software landscape is crucial to support scientists in their every day work, we recognized during our work with scientific institutes that many simulation programs can be considered legacy monolithic applications. They are developed without adhering to known software engineering guidelines, lack an acceptable software ergonomics, run sequentially on single workstations and require tedious manual tasks. We are convinced that SOA concepts and the service composition technology can help to improve this situation. In this paper we report on the results of our work on the service- and service composition-based re-engineering of a legacy scientific application for the simulation of the ageing process in copper-alloyed. The underlying general concept for a distributed, service-based simulation infrastructure is also applicable to other scenarios. Core of the infrastructure is a resource manager that steers server work load and handles simulation data.

Bridging the Gap between Business and Scientific Workflows: Humans in the Loop of Scientific Workflows

Due to their different target applications business and scientific workflow systems provide different sets of features to their users. Significant amount of research is currently being done to employ the business workflow technology in the scientific domain. This usually means extending the workflow language and thus the modeling tool and execution engine. In this paper we aim to bring business and scientific workflows together in order to exploit the advantages of both. We explore the interplay between business and scientific workflows in the context of human interactions with the management of workflow execution. We present an approach and implementation based on BPEL and Pegasus and show that the approach can be beneficial to scientists.

Model-as-you-go: An Approach for an Advanced Infrastructure for Scientific Workflows

Most of the existing scientific workflow systems rely on proprietary concepts and workflow languages. We are convinced that the conventional workflow technology that is established in business scenarios for years is also beneficial for scientists and scientific applications. We are therefore working on a scientific workflow system based on business workflow concepts and technologies. The system offers advanced flexibility features to scientists in order to support them in creating workflows in an explorative manner and to increase robustness of scientific applications. We named the approach Model-as-you-go because it enables users to model and execute workflows in an iterative process that eventually results in a complete scientific workflow. In this paper, we present main ingredients of Model-as-you-go, show how existing workflow concepts have to be extended in order to cover the requirements of scientists, discuss the application of the concepts to BPEL, and introduce the current prototype of the system.

BPEL4Pegasus: Combining Business and Scientific Workflows

Business and scientific workflow management systems (WfMS) offer different features to their users because they are developed for different application areas with different requirements. Research is currently being done to extend business WfMSs by functionality that meets requirements of scientists and scientific applications. The idea is to bring the strengths of business WfMSs to e-Science. This means great effort in re-implementing features already offered by scientific WfMSs. In our work, we investigated another approach, namely combining business and scientific workflows and thus harnessing the advantages of both. We demonstrate a prototype that implements this idea with BPEL as business workflow language and Pegasus as scientific WfMS. Our motivation is the fact that the manual work to correctly install and configure Pegasus can be supervised by a BPEL workflow to minimize sources of failures and automate the overall process of scientific experimenting.

Process space-based scientific workflow enactment

In the scientific field, workflow technology is often employed to conduct computer simulations or computer supported experiments. The underlying IT infrastructure typically comprises resources distributed among different institutes and organisations all over the world. Traditionally, workflows are executed on a single machine while the invoked software is accessed remotely. This approach imposes many drawbacks which are outlined in this paper. To address these weaknesses, we investigate the application of decentralised workflow enactment in the scientific domain. In this context, we explore the employment of process spaces, a middleware for the decentralised execution of workflows. Furthermore, we propose the combination of process spaces with the concept of data references to increase the overall performance of distributed simulations based on workflows. The considerations are discussed with the help of a scenario that calculates and visualises the ink diffusion in water over a period of time.

Towards Simulation Workflows with BPEL: Deriving Missing Features from GriCoL

In this paper, we investigate the suitability of the generalpurpose workflow language BPEL to create executable simulation workflows. We therefore compare BPEL to GriCoL, a graphical language with proven applicability for simulation workflows in Grid environments. We discover a number of incomparable concepts in the two languages. On the one hand, BPEL’s unique features in comparison to GriCoL reveal the rationale behind the approach of using BPEL as basis for a simulation workflow language. On the other hand, based on the features of GriCoL, we are able to discuss how to extend BPEL in order to increase its expressiveness for simulation workflows.

Service-based integration of human users in workflow-driven scientific experiments

The use of information technology in research and practice leads to increased degree of automation of tasks and makes scientific experiments more efficient in terms of cost, speed, accuracy, and flexibility. Scientific workflows have proven useful for the automation of scientific computations. However, not all tasks of an experiment can be automated. Some decisions still need to be made by human users, for instance, how an automated system should proceed in an exceptional situation. To address the need for integration of human users in such automated systems, we propose the concept of Human Communication Flows, which specify best practices about how a scientific workflow can interact with a human user. We developed a human communication framework that implements Communication Flows in a pipes-and-filters architecture and supports both notifications and request-response interactions. Different Communication Services can be plugged into the framework to account for different communication capabilities of human users. We facilitate the use of Communication Flows within a scientific workflow by means of reusable workflow fragments implementing the interaction with the framework.

Views on Scientific Workflows

Workflows are becoming more and more important in e-Science due to the support they provide to scientists in computer simulations, experiments and calculations. Our experiences with workflows in this field and the literature show that scientific workflows consist of a large number of related information. This information is difficult to deal with in a single perspective and has changing importance to scientists in the different workflow lifecycle phases. In this paper we apply viewing techniques known from business process management to (service-based) scientific workflows to address these issues. We describe seven of the most relevant views and point out realization challenges. We argue that the selected views facilitate the handling of workflows to scientists and add further value to scientific workflow systems. An implementation of a subset of the views based on Web services and BPEL shows the feasibility of the approach. The presented work has the goal to increase additionally the acceptance of the workflow technology in e-Science.