|
|
 |
| |
|
|
|
Neural Network PID Control based Scheduling Mechanism for Cloud Computing |
|
|
|
PP: 789-796 |
|
|
|
Author(s) |
|
|
|
Xiaoqi Xing,
Bin Liu,
Dongyi Ling,
|
|
|
|
Abstract |
|
|
| With the progress of academia and industry, cloud computing is moving from theory to practice. Job scheduling is an
important issue in the high performance cloud computing environment. An appropriate scheduling mechanism can efficiently reduce
the response time and enhance the user experience. However, most of the traditional scheduling mechanisms focus on the response time
or the security, and the load balancing strategy is static, which cannot satisfy the new trend of development of cloud computing. Based
on the new features of multi-users and multi-tasks in the open computing environment, the paper proposes a fuzzy neural network PID
(Proportional+Integral+Derivative) control based scheduling mechanism. The scheduler takes deviation between the current Quality
of Service (QoS) of the system and the average QoS as input, and adjusts the task scheduling of different virtual machines (VM) in
the cloud environment by using multi-input and multi-output fuzzy neural network PID control. In the scheduling mechanism, we
evaluate the load status of the VMs in the cloud, and adopt the Neural Networks to automatically tune the jobs of VMs to get the best
performance QoS of the system. Simulation shows the scheduling mechanism will not only improve satisfaction of customers but also
achieve load balancing effectively. |
|
|
|
|
|
|
|
