RAPHA: Rate-Based Page Fault Handling Mechanism in Virtual Memory System

In virtual memory system, whenever a new task is dynamically launched, it suffers from transient overload that caused by excessive page fault handling. As transient overload state occurred, the QoS of multimedia application may be degraded because the deadline miss ratio is increased. In this paper, we present efficient overload control mechanism for virtual memory system called RAPHA(RAte-based Page fault HAndling). A significant feature of the RAPHA algorithm is resource reservation for kernel activity and page fault dispersion based on per process limit. The RAPHA algorithm is implemented in the Linux operating system and its performance measured. The results demonstrate RAPHA’s superior performance in supporting multimedia applications. Experiment result shows that RAPHA mechanism could achieve 10%-20 % reduction in deadline miss ratio and 50%-60 % reduction in average delay