अमूर्त

Efficient Cache Utilization of Disk Array Failures

K.Shamili,P.Golda Jeyasheeli

Cache memory plays an important role in computer Architecture. The cache memory is the smallest and fastest memory, which reduces the average access to the storage device i.e. the RAID device. The existing cache replacement algorithms work effectively in fault free mode, but with poor performance. When one disk in the disk array fails, the request to the faulty disk is directed to the other non-faulty disks in the RAID disk to reconstruct the lost data. This increases the number of requests to the non-faulty disk. This decreases the performance and reliability of the system. To improve the performance, Victim Disk First (VDF) cache replacement algorithm is used. This algorithm retains the blocks of faulty disk in the cache.ie the blocks of faulty disk are not selected for replacement. This reduces requests to the other non-faulty disk. The VDF-LRU and VDF-LFU cache replacement algorithms are employed to replace the blocks of the non-faulty disks when the cache is full. VDF-LRU efficiently improves the performance by not selecting the blocks that are faulty.