Research On HC Model Based On Distributed Cache Memcached And Data Elimination Algorithm

At this stage,with the maturing of cloud computing technology,distributed cache technology is also get more attention and favor,distributed cache technology can provides good feedback because of its high degree of literacy,rapid expansion,support for concurrency,rapid response and other advantages.Memcached,a distributed cache system that provides good feedback when it comes to centralized access,has been widely used on large,highly visited websites due to its ability to speed up access in dynamic Web applications.However the users' requests are large and the access requests are relatively concentrated.A large number of users' accesses generate the cached data with high heat levels,and the cached data with high heat levels have a certain probability after being "cooled down" period of time by the user requests access again,that is,some high-heat cached data will be repeatedly "heated." Distributed cache Memcached treats all cached data the same,and does not improve the response speed of the cached data which always be repeated access to the "heating" feature.This paper analyzes the related technologies of distributed cache system Memcached,and combines the concept of distributed cache Memcached and hot data.In order to improve the response performance of hot-spot cached data,an HC(Hot and Cold)Model based on Memcached is proposed based on the fact that the popularity of cached data.The HC Model consists of any two designated Memcached servers.Memcached responds the requests by 3Level data elimination algorithm which has the priority of” hot-spot cached data server,cold-spot cached data server,and normal Memcached server”.The hot-spot cached data server stores hot-spot cached data,and the cold-spot cached data server stores cold-spot cached data.When the number of cached data accesses increases sharply,such cached data becomes hot-cached data according to the hot-calculating formula.The hot-spot cached data in each normal node will be collected and stored in the hot-spot cached data server.With the passage of time,the hot-spot cached data will be reduced in popularity,and the hot-spot cached data will be eliminated to the cold-spot cached data server according to the HCClockPro algorithm.The cold-spot cached data server acts as a buffer pool to prepare for the cold-spot cached data would be heated up later,then this part can return to the hot-spot cached data server.Finally,we verify the Memcached of HC(Hot and Cold)Model through simulation experiments.The experimental results show that HC Model can improve the performance of Memcached when accessing the same cached data repeatedly in a certain period of time.