Font Size: a A A

The Design And Implementation Of The Performance Management System For Data Center Network

Posted on:2015-09-23Degree:MasterType:Thesis
Country:ChinaCandidate:Y K LiuFull Text:PDF
GTID:2308330479979328Subject:Computer technology
Abstract/Summary:PDF Full Text Request
With the rapid development of computer science and global informatization, data center network is changing profoundly in many fields, such as composition, structure, function, and scale, etc. Data center always consists of many networks, and network performance and status will greatly affect the operation of the data center. Base on the background, the network management method used now cannot meet the need of data center. How to deal with the problems, such as the common management of data center, the assessment of performance and stability of the key link, and other issues, and how to find the weakness of the function in key points will be the important issue.First of all, the dissertation introduces the relative theory of network performance and the network structure of data center. Furthermore, we study the data collection methods, network health assessment model and other issues. In the end we design and build the management system of data center network. The main work in the dissertation is as follow:(1) According to the character of data center network, we build the model called DCPMM that belongs to data center network and has layered management based on JMX/RMI. The model can realize the unified management and can bring the real-time performance of the network and fault view to the user. It has also great expansibility meantime.(2) Combined equipment throughput, packet loss rate and error rates, and other important performance indicators, we build the health model of network and equipment, which plays the important role in the optimization of large-scale network computing performance. In order to break the limit of traditional method, we design a new one that is based on Adaptive Concurrency Control that has curvature data changes. According to the changes of data, the method can adjust the time interval dynamically. Meantime we use a multi- threaded concurrency control technology, which greatly enhance the collection efficiency.(3) Based on the management model DCPMM introduced in chapter 3, we design and build the management system of data center network that called PMS. The system, using layered structure, ensures the effectiveness and the accuracy. In the environment of real data center network, we test the system and get great result.
Keywords/Search Tags:Data Center Network, Performance Management, Network Health, Adaptive Concurrency Control
PDF Full Text Request
Related items