Study On Data Center Network Switching Performance And Security

The rapid growth of cloud computing requires large-scale and extensible data centers as infrastructure network support facilities to provide storage and computing resources for massive network services.The deepening development of cloud computing technology depends on the data center network to a large extent.Therefore,the research of data center has become a hot topic nowadays.With the explosive growth of various online services and applications,the requirement of providing high bandwidth and low latency for data centers is getting higher and higher.This essentially requires data centers to increase the rate of data transmission and exchange.Although optical switching can improve the rate of data transmission compared with electrical switching,due to the particularity of light,data can not be cached,so the configuration of switching links in switches requires additional overhead.Therefore,data center optical switching has become a very important research topic to improve the performance of transmission network.In addition,with the development of large data technology,the value of data is increasing.At the same time,users are more concerned about the degree of data privacy.The data center stores a large number of important data,so it also puts forward higher requirements for network security performance.At present,more and more enterprises introduce mature wireless technology to manage the operation and maintenance of data centers.This brings convenience to managers,but also greatly increases the network security risk caused by RAP(Rogue Access Points).In this paper,we study two aspects of data center network switching performance and security performance respectively.In the aspect of network switching performance,the batch packet scheduling method commonly used in optical switches is studied.Existing algorithms use matrix decomposition technology to decompose the traffic matrix into a set of permutation matrices with the same weights.This rigidity restricts the flexibility of scheduling to a certain extent and leads to large time slot waste.In order to solve this problem,this paper proposes an UWS(Unequal Weight Sequence)algorithm based on unequal weighted sequence,which can decompose the traffic matrix more flexibly and finely to ensure the performance of packet-loss-free transmission with limited packet delay.In terms of network security performance,we propose a wireless RAP location algorithm based on RSS(Received Signal Strength),which is suitable for data center.Firstly,the initial location of RAP is obtained by Distance-based localization method,and then a simplified Fingerprint-based method is adopted to improve the location precision.Our algorithm also gives the robust estimation interval of the localization effect on transmit power.Finally,we verify the feasibility and effectiveness of these two algorithms through a large number of experiments,and provide a new technical scheme for the safe and stable operation of the data center network.