Key Technologies Research Of High-Concurrency Logistics Information Platform

With the rapid development of internet,logistics information platform based on web is more and more widely used. as a consequence,the concurrent visits of logistics information platform increases,so the concurrent performance becomes a key factor to evaluate logistics information platform and a bottleneck of logistics information platform.This thesis analyzes the major subsystems of logistics information platform and proposes optimization plans for each subsystem.In database,based on Mysql,first,aiming at the shortage of the implementation of transaction in Mysql,this thesis proposes a optimistic lock based model and a pessimistic lock based model to avoid update lose after analzing the common scences of transaction using in logisitcis information platform.Then,this thesis does research on the foundation of table and index and proposes some general principles about the design of table schema and the design of index.Then, For scalability of database,this thesis raises a vertical-sharding pattern solution and a horizontal-sharding pattern solution to sacle out database.Then this thesis does research on cach system,gives a comparison between local cache and distributed cache and proposes a finer-grained and active-update cache solution for associated data in database.In terms of web server tunning, factors to evaluate the performance of web server are introduced. Based on JavaEE, this thesis does research on the JVM memory model and GC mechanism and proposes a general tunning method to JVM.Besides,based on tomcat,two key factors--threads amount and I/O model--that affect the performance of JavaEE web server are be studied. In the aspect of cluster of web servers,this thesis gives a comparison of different load balance technique and raises three solutions to solve the state-synchronization between web servers.at the end of this thesis,procedures of tunning of two practical web systems and the achievements of tunning is introduced.