Research On The Optimization Method Of Dynamic Website Cache Based On Recurrent Neural Network

Represented by cloud computing,social networking is the advent of the era of big data,makes the traditional business information needs continue to expand and network technology continues to advance at the same time,especially the mobile Internet technology,the function of the large-scale network applications and operating mode are also great changes have taken place,the site of concurrent traffic and exploding data processing scale,to the website system programs and data server load capacity and stability of the proposed challenge,at the same time of the system software and hardware performance put forward higher requirements.Caching technology is a key technology of large web site performance optimization,based on the performance of web cache optimization plan was studied,the main work includes:(1)Analyses the data cache server,reading and writing data cache data model is put forward,analysis the operation principle of cached in the model,and combined with the source code it gives the design and implementation of the scheme;aiming at the problems of cache file management and data retrieval,a buffer system architecture based on Lucene full-text search is proposed.The Lucene full-text search engine toolkit is used to realize the management and fast retrieval of massive cache files in the cache system.On the one hand,it reduces the use of memory space by the cache system,reduces resource consumption,and also reduces the frequency of cache updates in the memory space,so that only fixed types of local cache data are retained in the content.On the other hand,Lucene can be used to manage the cache files in the external storage space.It does not need to consider the limitation of the cache space,and does not need to restrict the cache data type and data volume.Instead,it only needs to use the corresponding data index to realize the retrieval and management of the cache files;(3)Aiming at the page caching problem of the server,a caching scheme based on cyclic neural network is proposed.In this scenario,the first in the source code execution level page with information about data collection,and then use cycle memory characteristic of neural network,identify the cache sequence patterns from these information,so that they can better choose the subsequent higher page or data cache access frequency,compared with the traditional cache elimination and selectionalgorithm,because the loop neural network considering the time and order page access,so the effect is better.The test and actual running results show that the proposed caching scheme can effectively improve the website performance and load capacity without affecting the website reliability.