Research And Implementation Of Key Technologies For Large Scale Data Center Network Performance Evaluation

Data center network is the infrastructure of large scale data processing and storage,which plays an important role in the application of Internet of things and cloud computing.With the increasing size of data center network,network failure rate is on the rise,which leads to the high cost of network operation and maintenance.Kinds of complex applications running on the data center,there are many network architectures,so the network traffic behavior is complicated.This situation Increases the difficulty of network management,and reduces the network resource utilization.Combined with historical data and network topology structure,this paper analyzes the impact of the data center network performance and network failures caused by various factors.To evaluate and warn early the equipment and network,the paper can effectively achieve the network optimization,and meet the needs of the data center network performance management.The main innovations of the paper are as follows:1.In view of the many parameters that affect the operation of network equipment,the overall performance of the equipment is more difficult to accurately assess the problem.To use quantitative index to reflect the performance of the equipment correctly,a method for evaluating the health degree of network equipment based on objective weight,and the dynamic adjustment mechanism of the weight of equipment performance parameters is used to calculate the health degree of network equipment.2.Aiming at the problem that the data center network is large and complex,it is difficult to optimize the performance of the network.A hierarchical network health degree evaluation method is proposed,which is based on the network topology,network level,priority,connectivity,and other factors.Use this method for network performance evaluation,it can reflect the operation of the whole network more accurately.3.There are problems in the process of operation of the equipment and network,such as performance degradation,severe failure and other abnormal conditions of slow response.A dynamic threshold based on sub health warning method is proposed,which is used to solve these problems.Through the analysis of the overall level of the devices and the network's health value and the fluctuation,the unhealthy network devices are screened out,and the sub healthy devices can be found out.The method also realize the retrospective function to figure out the reasons causing the network performance degradation.Then Provide more information for administrators.4.Using hierarchical management architecture design,the network performance evaluation management system is developed.From top to bottom,the system is composed of visual layer,performance management layer and data acquisition layer.from top to bottom,in turn for the visualization layer,performance management layer and data acquisition layer.It realizes a series of functions,such as the network data collection,data analysis,equipment and network health degree evaluation and the network early warning and etc,and and provides the friendly graphical user interface.The system has been deployed in Guangzhou Tianhe2 network,and is running in good condition.