In general, cost optimization problem for workflow
applications scheduling with deadline constraints is NPhard
[7]. A sub-optimal solution can be obtained by
local optimization strategies. Workflow execution may
expend much more cost at its earlier scheduling,
whereas the given deadline is usually greater than its
earliest completion time. Thus, releasing its finish time
to the deadline may decrease workflow execution cost.
The overall deadline is divided into task deadlines
without destroying their precedence constraints. Each
activity can select the cheapest service from its service
pool without exceeding its deadline. Local cost
optimization solution for each task can lead to the
minimum cost for the whole workflow execution and
meet the deadline demands. Therefore, it is crucial to
divide the whole deadline into task deadlines. Methods
are given in the following