当前位置: X-MOL 学术J. Cloud Comp. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Fine-grained load balancing with traffic-aware rerouting in datacenter networks
Journal of Cloud Computing ( IF 3.7 ) Pub Date : 2021-07-06 , DOI: 10.1186/s13677-021-00252-8
Tao Zhang 1, 2 , Yasi Lei 1 , Qianqiang Zhang 1 , Shaojun Zou 1, 2 , Juan Huang 1 , Fangmin Li 1, 2
Affiliation  

Modern datacenters provide a wide variety of application services, which generate a mix of delay-sensitive short flows and throughput-oriented long flows, transmitting in the multi-path datacenter network. Though the existing load balancing designs successfully make full use of available parallel paths and attain high bisection network bandwidth, they reroute flows regardless of their dissimilar performance requirements. The short flows suffer from the problems of large queuing delay and packet reordering, while the long flows fail to obtain high throughput due to low link utilization and packet reordering. To address these inefficiency, we design a fine-grained load balancing scheme, namely TR (Traffic-aware Rerouting), which identifies flow types and executes flexible and traffic-aware rerouting to balance the performances of both short and long flows. Besides, to avoid packet reordering, TR leverages the reverse ACKs to estimate the switch-to-switch delay, thus excluding paths that potentially cause packet reordering. Moreover, TR is only deployed on the switch without any modification on end-hosts. The experimental results of large-scale NS2 simulations show that TR reduces the average and tail flow completion time for short flows by up to 60% and 80%, as well as provides up to 3.02x gain in throughput of long flows compared to the state-of-the-art load balancing schemes.

中文翻译:

数据中心网络中具有流量感知重新路由的细粒度负载平衡

现代数据中心提供各种各样的应用服务,这些应用服务产生了对延迟敏感的短流和面向吞吐量的长流的混合,在多路径数据中心网络中传输。尽管现有的负载平衡设计成功地充分利用了可用的并行路径并获得了高二分网络带宽,但它们重新路由流而不管它们的不同性能要求如何。短流存在排队延迟大和数据包重排序的问题,而长流由于链路利用率低和数据包重排序而无法获得高吞吐量。为了解决这些低效率问题,我们设计了一种细粒度的负载平衡方案,即 TR(流量感知重新路由),它识别流类型并执行灵活和流量感知的重新路由,以平衡短流和长流的性能。此外,为了避免数据包重新排序,TR 利用反向 ACK 来估计交换机到交换机的延迟,从而排除可能导致数据包重新排序的路径。而且,TR只部署在交换机上,终端主机上不做任何修改。大规模 NS2 模拟的实验结果表明,TR 将短流的平均和尾流完成时间减少了 60% 和 80%,并且与 state 相比,长流的吞吐量提高了 3.02 倍最先进的负载平衡方案。TR只部署在交换机上,终端主机不做任何修改。大规模 NS2 模拟的实验结果表明,TR 将短流的平均和尾流完成时间减少了 60% 和 80%,并且与 state 相比,长流的吞吐量提高了 3.02 倍最先进的负载平衡方案。TR只部署在交换机上,终端主机不做任何修改。大规模 NS2 模拟的实验结果表明,TR 将短流的平均和尾流完成时间减少了 60% 和 80%,并且与 state 相比,长流的吞吐量提高了 3.02 倍最先进的负载平衡方案。
更新日期:2021-07-07
down
wechat
bug