ELAB：基于端系统的新型拥塞感知负载均衡机制

doi:10.11959/j.issn.1000-436x.2019054

摘要/Abstract

摘要：

良好的负载均衡机制是有效利用数据中心网络带宽的必备条件。现有的等价多路径路由（ECMP）的负载均衡方法由于负载均衡粒度过粗，且不具备对路径拥塞状态感知的能力，因此负载均衡效率较低。为解决此问题，近年来出现了一系列细粒度且具备拥塞感知能力的负载均衡研究工作。然而，这些研究或者需要修改交换机硬件以实时搜集网络各部位的拥塞情况，难以部署；或者虽不需要修改交换机硬件，仅需对端系统的软件进行修改，但由于缺乏准确的网络拥塞信息而导致负载混合效果不佳。针对该问题，提出了一种实现于端系统上的软件解决方案ELAB，该方案不需要对网络中的硬件进行修改，就可以达到良好的负载均衡效果。ELAB创造性地采用了基于可用带宽的方式进行流量负载均衡，相比于现有的基于端系统的方法，ELAB性能提升达20%。

关键词: 数据中心网络, 负载均衡, 基于边缘, 拥塞感知

Abstract:

A good load balance mechanism is the key to effectively use the network of the data center network.In current production data center,ECMP is the de facto load balancing scheme.However,it has two drawbacks.1) the load balance unit is too coarse-grained,2) it’s not congestion aware.To solve these problems,several fine-grained and congestion-aware load balancing works have emerged in recent years.These works either need to modify the switch hardware to collect congestion in various parts of the network in real time,and it is difficult to deploy; or only need to modify the end system,but the inaccurate sense of congestion cannot achieve a good load balancing effect.A novel edge-based load balancing scheme ELAB was proposed,which addresses above existing problems and improves the network performance up to 20%.

Key words: data center network, load balancing, edge-based, congestion aware

中图分类号:

TP393.1

陈果,张潍丰. ELAB：基于端系统的新型拥塞感知负载均衡机制[J]. 通信学报, 2019, 40(3): 196-205.

Guo CHEN,Weifeng ZHANG. ELAB:end-host-based congestion aware load balancing for data center network[J]. Journal on Communications, 2019, 40(3): 196-205.

图/表 13

图1

图2

图3

图4

图5

图6

图7

图8

图9

图10

图11

图12

图13

参考文献 20

[1]	BENSON T , AKELLA A , MALTZ D A . Network traffic characteristics of data centers in the wild[C]// The 10th ACM SIGCOMM Conference on Internet Measurement. 2010: 267-280.
[2]	SINGH A , ONG J , AGARWAL A . Jupiter rising:a decade of clos topologies and centralized control in google’s datacenter network[J]. ACM SIGCOMM Computer Communication Review, 2015,45(4): 183-197.
[3]	HOPPS C E . Analysis of an equal-cost multi-path algorithm[R]. Request for Comments 2992, 2000.
[4]	KATTA N , HIRA M , KIM C . Hula:scalable load balancing using programmable data planes[C]// The Symposium on SDN Research. 2016:10.
[5]	GHORBANI S , GODFREY B , GANJALI Y . Micro load balancing in data centers with DRILL[C]// The 14th ACM Workshop on Hot Topics in Networks. ACM, 2015:17.
[6]	ALIZADEH M , EDSALL T , DHARMAPURIKAR S . CONGA:distributed congestion-aware load balancing for datacenters[J]. ACM SIGCOMM Computer Communication Review, 2014,44(4): 503-514.
[7]	KATTA N , HIRA M , GHAG A . CLOVE:how I learned to stop worrying about the core and love the edge[C]// The 15th ACM Workshop on Hot Topics in Networks. 2016: 155-161.
[8]	ZHANG H , ZHANG J , BAI W . Resilient datacenter load balancing in the wild[C]// The Conference of the ACM Special Interest Group on Data Communication. 2017: 253-266.
[9]	VANINI E , PAN R , ALIZADEH M . Let it flow:resilient asymmetric load balancing with flowlet switching[C]// USENIX Symposium on Networked Systems Design and Implementation. 2017: 407-420.
[10]	HE K , ROZNER E , AGARWAL K ,et al. Presto:edge-based load balancing for fast datacenter networks[J]. ACM SIGCOMM Computer Communication Review, 2015,45(4): 465-478.
[11]	CAO J , XIA R , YANG P . Per-packet load-balanced,low-latency routing for clos-based data center networks[C]// The 9th ACM Conference on Emerging Networking Experiments and Technologies. 2013: 49-60.
[12]	DITTMANN G , HERKERSDORF A . Network processor load balancing for high-speed links[C]// The 2002 International Symposium on Performance Evaluation of Computer and Telecommunication Systems. 2002,735.
[13]	AL-FARES M , RADHAKRISHNAN S , RAGHAVAN B ,et al. Hedera:Dynamic flow scheduling for data center networks[J]. USENIX Symposium on Networked Systems Design and Implementation, 2010,10:19.
[14]	BENSON T , ANAND A , AKELLA A . MicroTE:fine grained traffic engineering for data centers[C]// The 17th Conference on emerging Networking Experiments and Technologies. 2011:8.
[15]	CURTIS A R , KIM W , YALAGANDULA P . Mahout:low-overhead datacenter traffic management using end-host-based elephant detection[C]// INFOCOM. 2011: 1629-1637.
[16]	PERRY J , BALAKRISHNAN H , SHAH D . Flowtune:flowlet control for datacenter networks[C]// USENIX Symposium on Networked Systems Design and Implementation. 2017: 421-435.
[17]	MAHALINGAM M , DUTT D , DUDA K . Virtual extensible local area network (VXLAN):a framework for overlaying virtualized layer 2 networks over layer 3 networks[R]. Request for Comments, 7348
[18]	KIM Y J , KOLESNIKOV V , KIM H . SSTP:a scalable and secure transport protocol for smart grid data collection[C]// 2011 IEEE International Conference on Smart Grid Communications (SmartGridComm). 2011: 161-166.
[19]	AUGUSTIN B , CUVELLIER X , ORGOGOZO B ,et al. Avoiding traceroute anomalies with Paris traceroute[C]// The 6th ACM SIGCOMM conference on Internet measurement. 2006: 153-158.
[20]	ISSARIYAKUL T , HOSSAIN E . Introduction to network simulator NS2[M]. Springer Science ＆ Business Media, 2011.