On Efficient Resource Use for Scientific Workflows in Clouds

0
557
On Efficient Resource Use for Scientific Workflows in Clouds

On Efficient Resource Use for Scientific Workflows in Clouds

Abstract

The abundance of cloud resources has enabled not only web applications but also scientific applications to easily scale to meet their goals, such as performance and cost. However, the decision on such scaling (resource management) is very complicated, often resulting in inefficient use of resources, due to the complex and large-scale nature of scientific workflows. In this On Efficient Resource Use for Scientific Workflows in Clouds paper, we present RDAS+ as a resource-sensitive scheduling algorithm to optimize resource efficiency to perform scientific workflows in clouds.

On Efficient Resource Use for Scientific Workflows in Clouds RDAS+ maximizes resource utilization by allocating the minimum number of resources (virtual machines or cloud VMs) with little time-consuming sacrifice (makespan). This optimization eventually leads to cost efficiency on pay-per-use cloud resources. RDAS+ consists of partitioning steps, resource allocation and task scheduling to realize such optimization. In comparison with three existing algorithms, we evaluated RDAS+ using five types of real-world scientific workflows.

System Configuration

H/W System Configuration
Speed                   : 1.1 GHz
 
RAM                      : 256 MB(min)
 
Hard Disk              : 20 GB
 
Floppy Drive          : 1.44 MB
 
Key Board             : Standard Windows Keyboard
 
Mouse                  : Two or Three Button Mouse
 
Monitor                : SVGA
 
S/W System Configuration
 
 
Platform                     :  cloud computing

 
Operating system       : Windows Xp,7,
 
Server                       : WAMP/Apache
 
Working on                : Browser Like Firefox, IE

Conclusion

In this paper, we summarized research issues for scientific workflow systems in cloud data management. We drew three directions for research that are data storage, data placement and data replication. We analyzed the existing research problems briefly in each direction, introduced promising methodologies and summarized state-of – the-art approaches.