A predictive adaptive load balancing model

Performance of load balancing scheduling policies in Web server cluster systems is greatly impacted by the characteristics of workload. Based on the analysis of the load characteristics for scheduling algorithm, a prediction-based adaptive load balancing model (RR_MMMCS-A-P) is proposed in this paper. Monitoring the workload characteristics and its variation, the arrival rate and the size of the follow-up request are predicted by RR_MMMCS-A-P and rapid adjustment of the corresponding parameters to balance the load between servers. Experiments have shown that compared with CPU-based and CPU-memory based scheduling strategy, RR_MMMCS-A-P have better performance in reducing average response time for both calculation-intensive and data-intensive jobs.

[1]  Mohammad Kazem Akbari,et al.  A predictive and probabilistic load-balancing algorithm for cluster-based web servers , 2011, Appl. Soft Comput..

[2]  Wei Sun,et al.  Workload-aware load balancing for clustered Web servers , 2005, IEEE Transactions on Parallel and Distributed Systems.

[3]  Li Xiao,et al.  Effective load sharing on heterogeneous networks of workstations , 2000, Proceedings 14th International Parallel and Distributed Processing Symposium. IPDPS 2000.

[4]  Wei Sun,et al.  ADAPTLOAD: effective balancing in clustered web servers under transient load conditions , 2002, Proceedings 22nd International Conference on Distributed Computing Systems.

[5]  Qingyang Meng,et al.  A Dynamic Load Balancing Method Based on Stability Analysis , 2008, 2008 International Symposium on Computer Science and Computational Technology.

[6]  Liu Xinhua,et al.  An Distributed and Directed Clustering Algorithm Based on Load Balance for Wireless Sensor Network , 2009 .

[7]  Daniele Puccinelli,et al.  Lifetime Benefits through Load Balancing in Homogeneous Sensor Networks , 2009, 2009 IEEE Wireless Communications and Networking Conference.

[8]  Gianfranco Ciardo,et al.  EQUILOAD: a load balancing policy for clustered web servers , 2001, Perform. Evaluation.