Research On Cache Replacement Strategies

Cache is an effective means to compensate for the huge difference in speed between theCPU and the storage system. With the limit of costs and production technology, the gap ofcapacity between the cache and storage system has a significant increase. It is a efficient wayof reducing the CPU idle time and improving CPU utilization as well as enhancing computerperformance by developing the potential storage performance.After deeply studying the data types and data access patterns changes in cache, the rootcauses of traditional replacement algorithm was summed up. A new cache replacement policyto solve the deficiencies identified through the introduction of simplified IRR model as thekey of replacement to enhance the ability of capture high-frequency data, cancel the queuecapacity constraints, improving the adaptability of the different data access patterns.Based on the strategy according to different application environment, two new cachereplacement algorithms are proposed-Aging Algorithm and Adapter Clock algorithm. Thoughintroducing the age record data for historical information, the aging algorithm can enhance theability to capture high-frequency data. Meanwhile, the flexibility of the algorithm can beenhanced by adding the capacity queue. Adapter Clock absorbed the idea of the Clockalgorithm, use the hand of the clock and flag record historical information of the date, whilesimplifying the organizational structure of the algorithm reduces the computational and thenumber of comparisons, the same competitive way to set up the collection capacity, andenhance the flexibility and adaptability of the algorithm.Certification program through the introduction of a cycle sequencing technology as aninput, shielding the impact of the computer’s performance, the certification program focusedon the algorithm itself. The test results are eventually shown in the form of a comparison chart,more intuitive and image demonstration of the performance of the new strategy. Theexperimental results show that: compared with the traditional replacement algorithm, the newstrategy to enhance the ability to capture high-frequency data, and enhance the flexibility ofthe algorithm to improve the cache hit rate, which has a better application prospects.