Research On Management Method Of Cache Data In Web System

With the rapid development of the Internet,the number of Web system visits is increasing,network congestion and user access delays are becoming more serious.The existing network bandwidth has not been able to satisfy users' service requests.Therefore,data caching technology came into being Cache technology stores some data in server memory or client,which can solve the above problems well,but most common cache technologies only consider the management process for all cached data,and do not consider data with different access frequency in cached data.Object management process,resulting in low cache hit rates.Based on the above shortcomings,this paper proposes a cache management method based on hot data prediction for e-commerce systems with the highest concurrent access in web application systems.Firstly,by comparing multiple data prediction methods,a method for predicting hotspot data based on support vector machines is proposed.Secondly,a combination of prefetching and cache management is used in cache implementation to predict the hotspot data obtained by prediction.Fetch and cache,non-hot data supports real-time caching.In terms of cache management,a method based on hot data and non-hot data is separately managed:(1)In terms of cache replacement,based on the high frequency of hot data access over a period of time,a replacement strategy for hot data is proposed to extend the cache resident time of hot data,respond to user requests faster,and increase the cache hit rate.(2)In terms of cache data consistency maintenance,based on the difference in the requirements of hot and non-hot data,collect server update methods and time-to-live policies to update hot and non-hot cache data.(3)In order to better ensure that the system has sufficient memory space at runtime,this paper proposes a cache space threshold strategy.The basic idea of the strategy is to give a threshold for the cache space,and then calculate the cache capacity by an algorithm.If the cache capacity exceeds the threshold,The expired cache data is cleared.Finally,a series of simulation experiments are designed to verify the above functions,including accuracy analysis of hot spot data prediction,comparison of cache data replacement strategies and hit ratio analysis,comparison analysis of cache data consistency maintenance,analysis of cache space control strategies and systems Comparative analysis of performance.