Portal and Workflow in Grid Computing: From Application Integration to Service Integration - Grid Computing: Infrastructure, Service, and Applications - page 288

11.3.1.3 Workfl ow Engines

A workl ow typically needs to be enacted by a workl ow engine. A work-
l ow engine is dei ned by WfMC as “A software service or 'engine' that
provides the runtime execution environment for a process instance.” It
coordinates the invocation of components within workl ows, and manages
the entire invocation process including progress reporting, data transfer
between components and any other housekeeping required. As the BPEL
is now a de facto standard workl ow language, currently there are many
workl ow engines that support BPEL standard; for example, Oracle BPEL
Process Manager, Microsoft BiTalk, IBM WebSphere Business Integration
Server Foundation, IBM BPWS4J, ActiveBPEL Engine and ActiveWebl ow,
and OpenStorm Service Orchestrator.

11. 3.1.4 Workfl ow Systems

A workl ow system is dei ned by Wf MC formally as “a system that dei nes,
creates and manages the execution of workl ows through the use of soft-
ware, running on one or more workl ow engines, which is able to inter-
pret the process dei nition, interact with workl ow participants and, where
required, invoke the use of IT tools and applications.” The widely used,
open-source, and service-based workl ow management systems are
Kepler, Taverna, and Triana.
The Kepler system [1] focuses on many application areas from gene pro-
moter to mineral classii cation. It can orchestrate standard Web services
linked with both data and control dependencies, and implement various
execution strategies. To support the interaction with Web services, Kepler
uses a form of actor proxy for each Web service invoked. In addition, they
have created a set of grid actors for doing grid FTP management and
Globus GRAM execution [1]. Triana [57], from the GridLab project, pro-
vides composition and a large toolbox of ready-to-use components. Triana
implements parallel and peer-to-peer policies and has been applied to
various scientii c i elds. For grid applications, Triana provides a mecha-
nism to move subgraph to remote grid resources for execution. It also
serves as a programming environment, providing a user interface to
enable the composition of scientii c workl ow applications. The Taverna
from myGrid project [1] targets bioinformatics applications and is able to
enact Web services and other components. Taverna creates a workl ow
based on service composition and provides over 1000 services that can be
used as components in workl ows [1].

11. 3.2 Workflows in Grid Computing

Workl ows in grid computing have some new issues [59]. First, on com-
putational grids, workl ow applications are not only from a business