Design And Implementation Of Cloud Cache System Based On Memcached

As the application system becomes more and more complex,each time the system not only needs to deal with a large amount of data but also needs to respond quickly to requests.In order to solve this problem the cache technology can be considered,at the same time we can add the idea of cloud computing in the cache technology to improve the system load capacity and scalability,so the concept of cloud cache was put forward.Cloud caching is based on distributed technology and from a new perspective to design and implementation of the cache function.Distributed technology can be a good combination with the original server cluster.Not only can effectively improve the server response speed but also reduced duplication of queries from back-end databases.This thesis will systematically introduce the design and implementation of cloud cache system based on memcached.First of all,this thesis points out the traditional memcached problems in the practical application environment.Memcached is centralized,when system faces a large number of concurrent requests not only can't do load balancing but also there will be a single point of failure,memcached does not have a user authentication mechanism,user legitimacy can't be guaranteed in the system.In view of the above problem,we put forward the multi-tenant,the high availability of the service,the high reliability of the data and the ease of system demand for the cloud cache system.According to the demand,the overall structure of the cloud cache system is put forward and the detailed design is made,the entire system will be divided into four types of nodes,including access nodes,management nodes,data nodes and maintenance nodes.Secondly,the function modules are designed and implemented in detail.Access nodes need to authenticate users who use cloud caching,and distribute the unique prefix to each user to isolate the user's data,using LVS technology combined with Reactor mode to handle a large number of concurrent requests;The management node controls the activity information of the whole system,and establishes the data directory index based on the B + tree to improve the data query speed,using the HWF(highest priority method)algorithm to select the data node server for data storage,the data stored in the form of multiple copies to ensure high reliability of the data,using Master / Slave model and MESI protocol to achieve multi-copy data consistency and data read and write separation;The data node is responsible for storing the data of the cache,using a fragmented manner to allocates and reclaims the storage space;The maintenance node is responsible for recording the operation log and running status of the whole system,providing some reference for the system optimization,in the case of abnormal system,the maintenance node can automatically detect and repair the exception,so that the system has a certain degree of self-repair.Finally,the cloud cache system is deployed in the actual environment,and the function and performance of the system are tested.The results show that the system completed the initial design.In the performance test,the system has a good I / O capabilities,and can do a quick response to request.