A Scheduling Framework for Web Server Clusters with IntensiveDynamic Content Processing

Clustering support with a single-system image view for large-scale Web serversis important to improve the system scalability in processing a large number ofconcurrent requests from Internet, especially when dynamic content generationusing CGI or other protocols becomes increasingly popular. This paper studiesa two-level scheduling framework with a master/slave architecture forclustering Web servers. Such an architecture has advantages in dynamicresource recruitment, fail-over management and it can also improve serverperformance compared to a flat architecture. The key methods we propose tomake this architecture efficient are the separation of static and dynamiccontent processing, low overhead remote execution, and reservation-basedscheduling which considers both I/O and CPU utilization. This paper provides acomparison of several scheduling approaches using experimental evaluation andanalytic modeling. The results show that proper optimization in resourcemanagement can lead to over 65% performance improvement for a fixed number ofnodes, and can achieve more substantial improvement when considering idleresource recruitment.