Research And Implementation Of Cache Optimization Techologies Based On Group For RAID

With the rapid development of the information technology, data storage is more and more emphasized. The speed gap between data processing in CPU and data accessing in storage system becomes more and more great,disk I/O has become a major factor of influncing performance of the computer system. In order to compensate for such gap in speed, Cache has been added in RAID(Redundant Array of Independent Disks). Cache could improve I/O performance of the storage system by reducing overhead of disk access and decreasing I/O response time.So now,how to reasonably organize and manage cache for maximizing the effectiveness and making the whole computer system running fast and efficient,has been a hot research topic in storage industry.The key to enhance the performance of cache is to improve the hit rate and reduce the I/O time of the page replacement.Then rational organization and scheduling strategies for cache is the key to improve hit rateand search efficiency. Using radix tree cache could realize rapid positioning,which reduce the search time greatly.And on the basis of the algorithms of Least Recently Used (LRU) ,we improve the replacement strategy using cache group strategy according to principles of spatial locality, and optimize the prefetching strategy to improve LRU algorithm.Caches of adjacent address make up a dynamic group, So Cache hit rate can be improved ,and I/O time can be reduced to further improve data access speed.Principles of cache grouping is that all cache operations are implemented on the group.If cache block was hitted,in accordance with the principle of spatial locality,the other caches in this group ,closed with this hitted cache,may be accessed in the future,so all caches within the group will be moved to the front of cache queue.This can improve the cache hit ratio,on the other hand can maintenance the continuity of cache address.When Caches will be writed back as group,this can effectively reduce the disk seek and rotationtime of physical head,thereby reducing the I/O time.At the same time,data access for multiple users will display the randomness,through the group searching ,it can effectively search the sequence of data access for each user.Using Iometer to test this system the results show that group cache optimization could improve the Cache hit rate and IO access speed.