Design And Implementation Of A Cache System Based On Association Rules

With the rapid development of information technology,diverse data have been collected and accumulated in various fields and industries.Information has become particularly important to every industry in society and users are increasingly demanding services with higher quality.In paticular,users are paying more attention to the relevance and response time of information acquisition.In the face of massive and disorderly data,finding an effective data processing method to deal with the needs of users,is of great importance and value for researches.In this paper,the association rules in data mining are applied to the storage system,and the optimized cache management scheme enables the cache management to run more efficiently based on the data association relationship.Through analysing of the history records of data access,the data association rule is obtained to optimize the data cache management,so that the system cache management can find the corresponding data association rules when reading the requested data,if the requested data meets the pre-conditions of the data association rules,the data associated with the request data will also be read into the cache to achieve the purpose of improving system I/O performance.We designed and implemented prototype of our system,which mainly includes a global management module that manages configuration of other module provides APIs of whole system;a data processing module that receives external data requests;a data analysis module that analyzes the data access history in the cache and figures out association rules;cache management module manages the organizes data in the cache and optimize it based on the association rules which produced by data analysis module;and the persistent storage module is responsible for interaction with the underlying storage system.From the result of evaluation by using SBS and SHOP,the improvement of performance is up to 9.3% and 5.4% under the same confidence condition comparing with the traditional solution,13.5% and 15% improvement of performance under the same support condition.