文章研究并解决数据中心的远程内存直接读取(remote direct memory access, RDMA)技术的网络拥塞控制问题。针对主流拥塞控制算法数据中心量化拥塞通知(data center quantized congestion notification, DCQCN)的收敛速度慢和缺乏硬件...文章研究并解决数据中心的远程内存直接读取(remote direct memory access, RDMA)技术的网络拥塞控制问题。针对主流拥塞控制算法数据中心量化拥塞通知(data center quantized congestion notification, DCQCN)的收敛速度慢和缺乏硬件实现方案的不足,提出可参数硬件化的数据中心量化拥塞通知(parameterized DCQCN,DCQCN-p)算法,该算法通过优化拥塞流的速度因子a、g调整速度比例Rc,并通过电路设计减少降速的频次;通过建立算法模型和搭建网络仿真NS-3平台,对比DCQCN-p算法在面临拥塞时单个调度流速度调整的性能以及多个调度流并发情况下的时延和吞吐量。仿真结果表明:在单个流面临拥塞时,DCQCN-p算法的数据传输速率比DCQCN算法的提高了50%;DCQCN-p算法在链路上最小速率为13.28 Gbit/s,相较于DCQCN、TIMELY、数据中心传输控制协议(data center transmission control protocol, DCTCP)算法,分别增长了24%、48%、23%;DCQCN-p算法(方差65%)的带宽分配公平性相较于TIMELY算法(方差216%)和DCTCP算法(方差191%)表现出显著的性能提升。展开更多
Cloud computing is becoming a key factor in the market day by day. Therefore, many companies are investing or going to invest in this sector for development of large data centers. These data centers not only consume m...Cloud computing is becoming a key factor in the market day by day. Therefore, many companies are investing or going to invest in this sector for development of large data centers. These data centers not only consume more energy but also produce greenhouse gases. Because of large amount of power consumption, data center providers go for different types of power generator to increase the profit margin which indirectly affects the environment. Several studies are carried out to reduce the power consumption of a data center. One of the techniques to reduce power consumption is virtualization. After several studies, it is stated that hardware plays a very important role. As the load increases, the power consumption of the CPU is also increased. Therefore, by extending the study of virtualization to reduce the power consumption, a hardware-based algorithm for virtual machine provisioning in a private cloud can significantly improve the performance by considering hardware as one of the important factors.展开更多
文摘文章研究并解决数据中心的远程内存直接读取(remote direct memory access, RDMA)技术的网络拥塞控制问题。针对主流拥塞控制算法数据中心量化拥塞通知(data center quantized congestion notification, DCQCN)的收敛速度慢和缺乏硬件实现方案的不足,提出可参数硬件化的数据中心量化拥塞通知(parameterized DCQCN,DCQCN-p)算法,该算法通过优化拥塞流的速度因子a、g调整速度比例Rc,并通过电路设计减少降速的频次;通过建立算法模型和搭建网络仿真NS-3平台,对比DCQCN-p算法在面临拥塞时单个调度流速度调整的性能以及多个调度流并发情况下的时延和吞吐量。仿真结果表明:在单个流面临拥塞时,DCQCN-p算法的数据传输速率比DCQCN算法的提高了50%;DCQCN-p算法在链路上最小速率为13.28 Gbit/s,相较于DCQCN、TIMELY、数据中心传输控制协议(data center transmission control protocol, DCTCP)算法,分别增长了24%、48%、23%;DCQCN-p算法(方差65%)的带宽分配公平性相较于TIMELY算法(方差216%)和DCTCP算法(方差191%)表现出显著的性能提升。
基金supported by the National Research Foundation (NRF) of Korea through contract N-14-NMIR06
文摘Cloud computing is becoming a key factor in the market day by day. Therefore, many companies are investing or going to invest in this sector for development of large data centers. These data centers not only consume more energy but also produce greenhouse gases. Because of large amount of power consumption, data center providers go for different types of power generator to increase the profit margin which indirectly affects the environment. Several studies are carried out to reduce the power consumption of a data center. One of the techniques to reduce power consumption is virtualization. After several studies, it is stated that hardware plays a very important role. As the load increases, the power consumption of the CPU is also increased. Therefore, by extending the study of virtualization to reduce the power consumption, a hardware-based algorithm for virtual machine provisioning in a private cloud can significantly improve the performance by considering hardware as one of the important factors.
