Xiao Liu PhD Thesis.pdf - Faculty of Information and Communication ...

to 5 which is normally large enough to compensate for time deficit with the given
compensation rate. The success rate for temporal violation handling is set as 80%.
CSS TD is applied with default values as defined in [22]. The fixed confidence
threshold for RA is set as 0.9 (i.e. select 10%) so as to be comparable to our strategy,
denoted as AD . The initial probability threshold and the update rate for our strategy
are set as 0.5 (mid-range initially) and 5% (a small percentile) respectively. This is a
moderate and reasonable setting when there is no prior knowledge. Additionally, to
reflect the effectiveness of temporal violation handling, the results of workflow
execution under the natural condition without temporal violation handling (denoted
as NIL ) are also presented.
Here, before we demonstrate the experiments on temporal violation handling
point selection in detail, we present an overview of the results on the cost for
handling temporal violations in scientific workflow systems. The two representative
temporal violation handling strategies investigated in our system are workflow
rescheduling and extra resource recruitment. Workflow rescheduling is to
compensate the time deficit by optimising the current activity-resource assignment.
The major overhead involved in workflow rescheduling includes activity transfer
(data and activity definitions) between resources, and the computation overhead of
the rescheduling strategy itself. In our SwinDeW-C scientific workflow system, for
a typical workflow rescheduling scenario (with around 200 activities and 10
resources), the average overhead for activity transfer is around 2 minutes and the
computation overhead is around 0.2 minutes [56]. As for extra resource recruitment,
it is to compensate the time deficit by employing additional resources for the
violated workflow instances at runtime. Its major overhead includes the set-up time
for new resources and the task transfer time. For a typical resource recruitment
scenario (adding one new resource), the normal set-up time for a single resource is
several minutes (like the Amazon EC2 Reserved Instances:
http://aws.amazon.com/ec2/reserved-instances/) and the activity transfer time (for
fewer than 5 activities in a local workflow segment) is around 0.3 minutes.
Furthermore, to give a direct view for the savings on the total temporal violation
handling cost, we record the reduced overhead for the 10 test cases in each round of
the experiment. For example, if we specify the basic time unit as one second, the
results have shown that for a large scale scientific workflow (with over 10,000
114