[Back] [PDF] [PS]

Dynamic Load balancing for I/O- and Memory- Intensive workload in Clusters using a
Feedback Control Mechanism

Xiao Qin, Hong Jiang, Yifeng Zhu, and David R. Swanson

Department of Computer Science and Engineering
University of Nebraska-Lincoln
Lincoln, NE 68588-0115, {xqin, jiang, yzhu, dswanson}@cse.unl.edu 

One common assumption of the existing models of load balancing is that the weights of resources and I/O buffer size are statically configured. Though the static configuration of these parameters performs well in a cluster where the workload can be predicted, its performance is poor in dynamic systems where the workload is unknown. In this paper, a new feedback control mechanism is proposed to improve the overall performance of a cluster with I/O-intensive and memory-intensive workload. The mechanism dynamically adjusts the resource weights as well as the I/O buffer size. Results from a trace-driven simulation show that this mechanism is effective in enhancing the performance of a number of existing load-balancing schemes.  

in Proceedings of the 9th International Euro-Par Conference on Parallel Processing (Euro-Par 2003),
Klagenfurt, Austria, August 26- 29, 2003.