The Optimization Research Of Traffic Engineering Fordata Center Networks Based On ECMP

With the rapid development of cloud computing, data center has been the core andinfrastructure of information facility. With the development of new technology and thepopularity of the service of cloud computing, significant changes are taking place indata centers and bring new challenges and problems to data center networks. Emergingservices needs a lot of communication between virtual machines or servers, the result isthe rapid growth of internal traffic within data center which shows different featuresfrom Internet traffic. When traditional data center network technology has to confrontwith the changes brought by cloud computing, network resource allocates inadequately,resulting in link congestion, high packet loss rate, large latency, which affect applicationperformance and quality of service.The main work of this paper includes the following four aspects: description of thedevelopment of data center network topology architecture and the data center networktechnology trend; research on traffic engineering and load balancing technology withindata center network under the cloud computing technology; against the typical ECMP(Equal Cost Multi Path) mathematical model,we optimize ECMP model by adding flowdistribution ratio parameter so that it can dynamically balancing the load; we design aheuristic traffic engineering optimization strategy based on end to end link utilizationand implement simulation of the optimization algorithm.First, we summarize the background and the latest progress of the data centernetwork, comprehensively introduce the current data center network architecture andnetwork technology development.Second, with the characteristics of data center traffic as the breakthrough point,from the view of the data center traffic engineering, through study the existing datacenter traffic engineering techniques and algorithms, find that the constraints of networkperformance and traffic engineering optimization is focused on the rational use ofequal-cost multipath.Third, against with ECMP technology, research on the ECMP mathematical modeland the optimized research, Due to the ECMP optimization mathematical model is NP(Non-deterministic Polynomial) problem, in order to solve this problem, we design aheuristic traffic engineering optimization strategy based on end to end link utilization.The optimization goal is end-to-end equal-cost path load balancing, optimization methods use simulated annealing algorithm to calculate the flow distribution ratiobetween equal-cost paths, reduce the high load equal-cost path ratio.Finally, this paper use NS-2simulation platform, through a lots of simulation tests,we verify the algorithm to obtain the optimal distribution ratio. Comparison with thetraditional simulation based on the HASH algorithm and the ideal model ofNon-blocking, the algorithm can be obtained better improve the whole network linkutilization, reducing end-to-end delay, reducing packet loss.