A Data Stream View of Scientific Workflow - Grid Computing: Infrastructure, Service, and Applications - page 226

9.2
As discussed previously, the scientii c workl ow can be treated as data
streams. Sometimes, the models of the scientii c workl ow can be described
as query expressions of XML data streams. In fact, in many research i elds,
workl ow models are also called as schemas [18,19]. The following section
gives the dei nition of a workl ow schema.

Stream View of Scientific Workflow Schema

9.2.1 Workflow Schema

A workl ow schema dei nes a state machine (deterministic i nite auto-
mation—DFA) [18,19], consisting of

• States, including a marked initial state

• Transitions

• State variables

Under the general scheme, the workl ow schema ( S ) is dei ned through
a directed graph consisting of nodes ( N ) and l ows ( F ) [14]. Flows show the
control l ow of the workl ow. Thus S = < N , F > is a directed graph where N
is a i nite set of nodes, F is a l ow relation F ⊆ N ¥ N . Nodes are classii ed

into tasks ( T ) and coordinators ( C ), where C T = f.

Task nodes represent atomic manual/automated activities or subpro-
cesses that must be performed to satisfy the underlying business process
objectives. Coordinator nodes allow us to build control l ow structures to
manage the coordination requirements of business processes. Basic schema
modeling patterns supported by these coordinators include sequence,
exclusive or-split (choice), exclusive or-join (merge), and-split (fork), and
and-join (synchronizer) [14]. An instance within the workl ow graph repre-
sents a particular case of the process. Figure 9.4 gives an example of the
graph-based workl ow schema.

∪ T , C ∩

B

Activity

Choice
split

Choice
merge

Pattern

Pattern

C

Activity

Parallel
split

Start A Synchronization G End

Graph-based view for
workflow schema

Pattern Activity

Activity Pattern

D E F

Activity Activity Activity

FIGURE 9.4 The workl ow schema with basic patterns.