The Realization And Performance Optimization Of The Instant Messaging System Based On Distributed Architecture

Along with the rapid development of Internet,network communication is playing an indispensable role in people's daily work and life.Network instant messing which has the advantages of a high speed,convenient and free,is increasingly popular with the public.Currently,many companies use or develop a safely independent internal instant messaging system to handle internal business communications.As users amounts huge increasing,a large number of user data and the expanding of service modules will put forward more challenges on backend servers.In this paper,we make a basic research of mass data storage and processing.On the basis of HBase distributed storage structure and data retrieval,we processes a data querying optimization algorithms which based on multilevel hash index tables.Also a server forwarding model is designed and implemented based on distributed structure of Servers,making instant messaging system a more efficient data analysis and processing performance which in the background of secondary user scale and large amount of data.Specific work is as follows:(1)The system structure,functions of components,data model and data transport model of the instant messaging system are designed and implemented.The enterprise data storage and analysis of the IM system processing are based on distributed architecture,which working in C/S communication mode and providing messaging service for both employees and item teams.(2)On the basis of analyzing the characteristics of enterprise distributed servers,we design a distributed architecture mode for tcp forwarding in servers,giving full play to advantages in more high-performance concurrent processing of distributed architecture under the server cluster,can satisfy needs in the moderate to large scale of users and data application.(3)A HBase optimized query algorithm is proposed based on multiple hash index tables for low efficiency when querying data under the situation of large scale of data storage.The algorithm first map HBase data which connected with time domain to 2D space objects,then use the K-Means clustering algorithm to divide time domain objects,finally build multistage HBase hash index tables according to the division of sub areas which contained in rectangles.The instant messaging system designed and implemented based on HBase and a multi-threading style,can effectively guarantee a high speed of data storage andprocessing in server cluster.As the branch of the enterprise informatization office system,the system not only provides instant communication services,but also connected with enterprise business closely,laying a good foundation for further expanding of system functions.