Introduction - User-Level Workflow Design: A Bioinformatics Perspective - page 7

Fig. 1.4 Exemplary image processing workflow model
described above: After selecting the picture for the birthday card, it is decided
if it needs to be rotated. If so, a rotate service (based on a corresponding
function of the ImageMagick tool convert ) is executed. If not, the workflow
directly proceeds to the next step, where normalize performs a transforma-
tion of the image so that is spans the full range of colors. Then, the custom
text is added ( draw ) and the picture set to a style that simulates a Polaroid
picture ( polaroid ), before the resulting birthday card is finally sent to the
recipient via e-mail.
Note that in contrast to frequently made reservations, the overhead that
is introduced by the model-driven development process does not compro-
mise the runtime performance of the workflows (cf., e.g., [175] or [142, Sec-
tion 5.3.2]): Many workflows orchestrate remote services, that is, the major
part of the involved computations is carried out transparently by a remote
system, and the workflow itself is mainly concerned with the interfaces of the
remote services and the “glue code” between them. Only a small part of the
computations takes place on the local machine, and the overall workflow per-
formance is hardly influenced by the local workflow execution performance.
In fact, also when the workflow execution is not distributed over a network
and no network delays impact the execution times, the majority of the com-
putations is still carried out within the individual services, which can not be
influenced by the workflow environment that simply calls them.
1.1.2 Challenges in Scientific Workflow Development
Dealing with the bioinformatics application domain, this topic focuses on sci-
entific workflows (cf., e.g., [37, 106, 355, 193, 258]), that is, combinations of
activities and computations in order to solve scientific problems. Scientific
workflows typically deal with complex, computationally intensive data analy-
sis procedures that are carried out in order to confirm or invalidate a scientific
hypothesis [193]. This is in contrast to, for instance, business processes im-
plement real-world procedures concerned with the handling of documents and