Pattern Based Real Time Disk Scheduling Algorithm for Virtualized Environment

Abstract

Achieving deadline in real time environment is a real practical difficulty. Real time scheduling algorithms consume more time in making scheduling decisions which leads to increase in deadline misses. The problem is further complicated in virtual environment. This motivates the idea for designing an efficient algorithm for future prediction of block access requests in virtual environment. Disk block access requests made by real time applications in virtualized environment are analyzed. Future disk access request are predicted by non-work conserving disk scheduling algorithm in offline mode. These pre-fetched disk blocks are moved to buffer cache. Pattern based detection technique is applied for predicting the future access of buffer cache blocks. Executing processes access large amount of data and require disk accesses. In real time virtualized environment the requesting processes are scheduled using an adaptive real time scheduling algorithm which reduces the deadline misses. Thus, due to non-work conserving algorithm seek time is reduced. Pattern based technique improves the hit ratio and reduces I/O time. An adaptive real time scheduling algorithm helps to schedule the processes for achieving their deadlines. Thus the performance of applications is enhanced in virtual environment and therefore non-work conserving pattern based adaptive real time scheduling algorithm is found very useful for hard real time applications.