Research And Implenmation On High Performance Network Fragment Data Processing System

As the Internet is widely used in different scenarios,the fragmentation and bursty data is also increasing.We need a protocol with less latency and less time consuming retransmission request to transfer the entire Internet,response and interaction.TCP is the underlying basis for many communication protocols,however,it needs its own connection time overhead.Tight congestion control algorithm slow start feature also brings a disadvantage in a highly concurrent small data transfer,which has become a difficult to eliminate defects.Massive fragmentation data transmitted to the receiver also brought great pressure on the service.Faced with massive fragmentation data generated from a variety of data sources,we need to use load balancing cluster technology to provide services lateral expansion capability.However,the existing load balancing techniques are mainly aimed to serve application-layer Web services,distributed computing,cloud platforms and are lack of support for data fragmentation scenarios.In response to these problems,this novel design and implement a high-performance network data processing system that provides efficient data transmission services for future network services.This paper carried out the following tasks:(1)Based on UDP design a data transfer protocol,remove the impact of the slow start;(2)Based on custom data transmission protocol design load balancing algorithm to provide the ability to scale the cluster level.The paper set up a mini test environment to functional verificate system and evaluate its performance.System modules are tested individually and the finally an overall test.Experimental results show that the system design features have been implemented,and in a highly concurrent connections short scene in contrast to the current system data throughput indicators have slightly improved.