Software-Defined Networking(SDN),with segregated data and control planes,provides faster data routing,stability,and enhanced quality metrics,such as throughput(Th),maximum available bandwidth(Bd(max)),data transfer(DT...Software-Defined Networking(SDN),with segregated data and control planes,provides faster data routing,stability,and enhanced quality metrics,such as throughput(Th),maximum available bandwidth(Bd(max)),data transfer(DTransfer),and reduction in end-to-end delay(D(E-E)).This paper explores the critical work of deploying SDN in large-scale Data Center Networks(DCNs)to enhance its Quality of Service(QoS)parameters,using logically distributed control configurations.There is a noticeable increase in Delay(E-E)when adopting SDN with a unified(single)control structure in big DCNs to handle Hypertext Transfer Protocol(HTTP)requests causing a reduction in network quality parameters(Bd(max),Th,DTransfer,D(E-E),etc.).This article examines the network performance in terms of quality matrices(bandwidth,throughput,data transfer,etc.),by establishing a large-scale SDN-based virtual network in the Mininet environment.The SDN network is simulated in three stages:(1)An SDN network with unitary controller-POX to manage the data traffic flow of the network without the server load management algorithm.(2)An SDN network with only one controller to manage the data traffic flow of the network with a server load management algorithm.(3)Deployment of SDN in proposed control arrangement(logically distributed controlled framework)with multiple controllers managing data traffic flow under the proposed Intelligent Sensing Server Load Management(ISSLM)algorithm.As a result of this approach,the network quality parameters in large-scale networks are enhanced.展开更多
As a kind of parallel storage system, RAID5 has been becoming a popular solution for providing better performance at low cost and without sacrificing much redundant data, its main disadvantage is poor performance. It ...As a kind of parallel storage system, RAID5 has been becoming a popular solution for providing better performance at low cost and without sacrificing much redundant data, its main disadvantage is poor performance. It is a general method to improve RAID5 performance by using cache. However, it often happens that the data is not hit in cache, in this case, the RAID5 performance also will be very poor. The method based on combination of disk I/O operations has been proposed for impriving the I/O response time through reducing the number of low-level operations. According to theoretic analysis and experimental test, we find that RAID5 access time and data transfer rate could be largely improved than conventional method.展开更多
文摘Software-Defined Networking(SDN),with segregated data and control planes,provides faster data routing,stability,and enhanced quality metrics,such as throughput(Th),maximum available bandwidth(Bd(max)),data transfer(DTransfer),and reduction in end-to-end delay(D(E-E)).This paper explores the critical work of deploying SDN in large-scale Data Center Networks(DCNs)to enhance its Quality of Service(QoS)parameters,using logically distributed control configurations.There is a noticeable increase in Delay(E-E)when adopting SDN with a unified(single)control structure in big DCNs to handle Hypertext Transfer Protocol(HTTP)requests causing a reduction in network quality parameters(Bd(max),Th,DTransfer,D(E-E),etc.).This article examines the network performance in terms of quality matrices(bandwidth,throughput,data transfer,etc.),by establishing a large-scale SDN-based virtual network in the Mininet environment.The SDN network is simulated in three stages:(1)An SDN network with unitary controller-POX to manage the data traffic flow of the network without the server load management algorithm.(2)An SDN network with only one controller to manage the data traffic flow of the network with a server load management algorithm.(3)Deployment of SDN in proposed control arrangement(logically distributed controlled framework)with multiple controllers managing data traffic flow under the proposed Intelligent Sensing Server Load Management(ISSLM)algorithm.As a result of this approach,the network quality parameters in large-scale networks are enhanced.
基金Supported by the National Natural Science Foundation of China under grant6 97730 46 and(6 99730 17) and Defense Advanced Resea
文摘As a kind of parallel storage system, RAID5 has been becoming a popular solution for providing better performance at low cost and without sacrificing much redundant data, its main disadvantage is poor performance. It is a general method to improve RAID5 performance by using cache. However, it often happens that the data is not hit in cache, in this case, the RAID5 performance also will be very poor. The method based on combination of disk I/O operations has been proposed for impriving the I/O response time through reducing the number of low-level operations. According to theoretic analysis and experimental test, we find that RAID5 access time and data transfer rate could be largely improved than conventional method.
