Dynamic Load Balancing for Web Clusters

The wide growth of Internet users has led the developers of popular Web sites to adopt the solution of substituting the Web servers by clusters of Web servers, in order to correspond efficiently to the high rate of requests received. They also applied load balancing techniques to distribute the workload among them. In this paper we present a general design of load balancing strategies in a cluster of Web servers, which uses a proxy per server to collect the incoming requests. To accomplish the best possible workload distribution, we use an update process, which can either be centralized or distributed. We apply two sender-initiated load balancing algorithms, based on the distinction between the centralized and the distributed update process, and analyze their performance results. We conclude that the distributed case behaves better under high loaded conditions.

[1]  Edward D. Lazowska,et al.  Adaptive load sharing in homogeneous distributed systems , 1986, IEEE Transactions on Software Engineering.

[2]  Azer Bestavros,et al.  Distributed packet rewriting and its application to scalable server architectures , 1998, Proceedings Sixth International Conference on Network Protocols (Cat. No.98TB100256).

[3]  Phillip Krueger,et al.  Two adaptive location policies for global scheduling algorithms , 1990, Proceedings.,10th International Conference on Distributed Computing Systems.

[4]  Azer Bestavros,et al.  Load balancing a cluster of web servers: using distributed packet rewriting , 2000, Conference Proceedings of the 2000 IEEE International Performance, Computing, and Communications Conference (Cat. No.00CH37086).

[5]  Tao Yang,et al.  Scheduling optimization for resource-intensive Web requests on server clusters , 1999, SPAA '99.

[6]  Oscar H. Ibarra,et al.  SWEB: towards a scalable World Wide Web server on multicomputers , 1996, Proceedings of International Conference on Parallel Processing.

[7]  Michelle Butler,et al.  A Scalable HTTP Server: The NCSA Prototype , 1994, Comput. Networks ISDN Syst..

[8]  Edward D. Lazowska,et al.  A Comparison of Receiver-Initiated and Sender-Initiated Adaptive Load Sharing , 1986, Perform. Evaluation.

[9]  Sivarama P. Dandamudi.  The Effect of Scheduling Discipline on Sender-Initia ted and Receiver-Initiated Adaptive Load Sharing in Homogeneous Distributed Systems† , 1995 .

[10]  Mark Crovella,et al.  Dynamic Server Selection using Bandwidth Probing in Wide-Area Networks , 1996 .

[11]  Paul G. Spirakis,et al.  A Competitive Symmetrical Transfer Policy for Load Sharing , 1998, Euro-Par.

[12]  Oscar H. Ibarra,et al.  Adaptive Partitioning and Scheduling for Enhancing WWW Application Performance , 1998, J. Parallel Distributed Comput..

[13]  Robert Martin McCool,et al.  Administering Very High Volume Internet Services , 1995, LISA.

[14]  Paul Barford,et al.  Generating representative Web workloads for network and server performance evaluation , 1998, SIGMETRICS '98/PERFORMANCE '98.

[15]  Tao Yang,et al.  Hierarchical resource management for Web server clusters with dynamic content , 1999, SIGMETRICS '99.