Information Technology Reference
In-Depth Information
In technical terms a workflow or, as it also is known, a workflow composition
means set of applications or entities connected to each other in order to process a
complex algorithm in cooperation. A workflow engine or enactor, is capable of
interpreting the workflow, identifying its nodes, and making decisions about which
nodes can be executed according to the data dependency and environmental cir-
cumstances. Workflow management systems (WfMS) consist of such interpreters
for workflow enactment and other additional tools that support the execution of the
workflow (e.g., handling data transfer transparently).
Research on workflow management belongs to the area of service compositions
(Dustdar 2005). This area can be divided into two widely known
fields concerning
the enactment type: service orchestration and service choreography (Peltz 2003).
Service orchestration de
nes the workflow enactment and thus decides which jobs
are executed according to the workflow structure. It may be done in an adaptive
way, namely taking the current state of the computational resources into consid-
eration. Workflow management is de
nitely based on this concept. In contrast,
Service choreography uses the idea of distributed enactment, where the enactment
decisions are made by the workflow nodes; hence bottleneck issues caused by the
single point of the enactor are resolved. To conclude, via WfMS tools scientists are
able to design, manage, and reuse their own experiments executed locally on their
own machine or by utilizing remote computational and/or storage facilities.
Widely known WfMS are Taverna (Hull 2006), Kepler (Altintas 2004; Luda-
scher 2006), Triana (Taylor_Triana 2007), Pegasus (Deelman 2005), ASKALON
(Fahringer 2007), and Galaxy (Goecks 2010; Blankenberg 2010; Taylor_Triana
2007). Moreover, scientists can share workflows as good examples with their
colleagues thanks to online workflow repositories such as MyExperiment (Goble
2007) or SHIWA Workflow Repository (Korhov 2011). By understanding the
importance of scienti
c research, WS-PGRADE/gUSE was
designed as a workflow-oriented science gateway framework where the most
essential component of the system is the workflow engine. All the other compo-
nents extend the workflow feature and are responsible for making its usage easier
and more convenient.
c workflows in scienti
3.2 Syntax of gUSE Work
fl
ows
Besides the term workflow, the other important term is the node (or job) that
represents one particular stand-alone entity of computation (executable, web-ser-
vice invocation, etc.). A workflow composition consists of connected nodes.
Ports represent data in many-to-one association with a node (namely many ports
can be added to a node, but every port must only be added to one node). Types of
ports can be set
“
”
“
”
denoting that the port represents a required input data,
or an expected output data. Ports can be connected to each other de
in
or
out
ning the
dataflow; thus, only ports of different types can be connected. Hence, this separa-
tion of ports de
nes an implicit semantics for the enactment of connected nodes.
Search WWH ::
Custom Search