Information Technology Reference
In-Depth Information
In addition cloud services are mostly expensive for computing, storage and band-
width, etc. The business model of pay as you go also can reduce the cost of workflow
execution [4, 5]. Therefore, optimization of the workflow execution in cloud compu-
ting has become one of the research hotspots in the research of workflow and cloud
computing recently.
Generally speaking, it is necessary to map tasks to the appropriate execution re-
source by some workflow scheduling algorithms in cloud, which will directly influ-
ence the success rate of cloud workflow scheduling and the execution efficiency.
Besides, unlike traditional workflow scheduling, cloud workflow scheduling should
consider not only the optimal combination and utilization of the resources but also the
constraint of time sequence and causation of each task to obtain the final result. As a
consequence, the cloud workflow scheduling problem is usually a NP-hard problem.
The implications of cloud workflow scheduling research are as follows:
•
It can promote the user's QoS request of gratification. It not only promotes the
user's gratification of workflow execution cost, but also attracts the users to use
cloud services, thus help to achieve maximum profit.
•
It can improve the resource utilization of cloud services provided by cloud service
provider. By taking the characteristic of workflow instance into account, the re-
source utilization involved in workflow scheduling will be significantly improved.
•
It can promote the development and application of cloud computing and workflow
technology, especially in the areas of biomedicine, chemistry, gene expression
data analysis, astrophysics and the instance-intensive applications such as
e-commerce, etc.
3
Typical Cloud Workflow Scheduling Algorithms
This section makes the typical cloud workflow scheduling algorithms with principle,
merit and demerit sorting and analysis according to the scheduling difference of existing
cloud workflows. We can divide the algorithms into three categories: single-objective
optimization algorithms, multi-objective optimization algorithms, and heuristic algo-
rithms for scheduling strategy and heuristic algorithms used.
3.1
Single-Objective Optimization Workflow Scheduling Algorithms
Cloud Workflow Scheduling Algorithm Oriented to Dynamic Price Changes
Min Zheng et al. [9]
proposed a cloud workflow algorithm based on dynamic planning
to solve the scheduling overhead optimization of cloud workflow in dynamic resource
prices environment. Firstly, they define the model of workflow, resource and the tar-
get of task. Then, the cloud workflow tasks are divided into groups. Next, using the
dynamic algorithm to dispatch each task and making the task links of workflow get
better result. After grouping, the overall deadline will be allocated to each task group,
and sort them topologically. At last, the dynamic algorithm is used to dispatch each
task, and that the lowest overhead scheduling scheme in certain time is calculated.
