SDN中基于流量聚合的局部快速重路由算法

93 浏览量更新于2024-08-29 收藏 838KB PDF 举报

本文档探讨了"Local Fast Reroute with Flow Aggregation in Software Defined Networks"这一主题，发表在2017年4月的IEEE COMMUNICATIONS LETTERS上。作者Xiaoning Zhang、Zijing Cheng、Rong Ping Lin、Lei He、Shui Yu和Hongbin Luo提出了一种新的局部快速重路由（Local Fast Reroute, LFR）算法，专为软件定义网络（Software Defined Network, SDN）设计。在传统的SDN架构中，当网络中的链路发生故障时，控制平面通常需要处理大量的流量重新路由操作，这可能导致控制器和交换机之间的通信负担加重，同时也可能延长恢复时间。LFR算法旨在解决这个问题。它的核心思想是，在链路故障检测到后，将所有受影响的流量流聚合为一个“大”流量，然后由SDN控制器动态地为这个聚合流量部署新的本地重路由路径。这样做的好处在于减少了控制器与交换机之间的流量操作次数，提高了网络恢复的效率。通过数值结果的展示，该研究证明了LFR能够在快速恢复网络的同时，有效地控制SDN中的流量表项数量，从而实现资源的有效利用和网络性能的优化。因此，LFR算法对于提高SDN的容错能力和整体管理效率具有重要意义。关键词包括：快速重路由、SDN、恢复和故障管理。这项研究对SDN的流量管理和网络恢复策略进行了深入探索，为构建更加健壮和高效的SDN架构提供了理论支持和技术方案。对于网络运营商和SDN技术开发者来说，理解并应用这种本地快速重路由算法有助于提升网络的可用性和响应速度，降低运营成本。

IEEE COMMUNICATIONS LETTERS, VOL. 21, NO. 4, APRIL 2017 785

Local Fast Reroute With Flow Aggregation in Software Deﬁned Networks

Xiaoning Zhang, Zijing Cheng, RongPing Lin, Lei He, Shui Yu, and Hongbin Luo

Abstract—In this letter, we propose a local fast reroute (LFR)

algorithm with ﬂow aggregation in software deﬁned net-

works (SDN). In LFR, if a link failure is detected, all trafﬁc

ﬂows affected by the failure are aggregated to be a new “big”

ﬂow. Then, a local reroute path is dynamically deployed by the

SDN controller for the aggregated ﬂow. LFR reduces the number

of ﬂow operations between the SDN controller and switches.

Numerical results show that the LFR enables fast recovery while

minimizing the total number of ﬂow entries in SDN.

Index Terms— Fast reroute, SDN, restoration, failure.

I. INTRODUCTION

OFTWARE Deﬁned Network (SDN) [1] is a new network

paradigm with control plane and data plane separated,

which allows operators to easily deploy network applications

through a central controller in the control plane, and to

distribute ﬁne-grained policies into switch ﬂow tables in the

data plane through standard interfaces. The central controller

that has a global network state is responsible for path selec-

tion, and it controls the network-wide distributed forwarding

devices with the OpenFlow protocol. Flexible structure, central

conﬁguration, and fast deployment of network services and

applications are amongst the beneﬁts of SDN.

Thanks to the considerable ﬂexibility offered by SDN,

resilience technologies against network failures may be very

efﬁcient. Resilience mechanisms in SDN are divided into two

categories: protection and restoration. In the case of protec-

tion [2], [3], backup path is always pre-planned and reserved

before failures. Hence, when a failure occurs, no additional

signaling is needed to establish protected paths, disrupted

trafﬁc ﬂows can immediately be redirected. Sharma et al. [2],

ﬁrst apply path protection in OpenFlow networks. When BFD

(Bidirectional Forwarding Detection) packets detect the failure

of a working path, the trafﬁc on the working path is switched to

the backup path according to the GroupTable that is speciﬁed

in OpenFlow v1.1. Sgambelluri et al. [3] proposed the segment

protection enabling fast recovery. Each link of a working

path is protected by a speciﬁed backup path. Fast recovery is

performed locally by the switch connected to the failed link.

Manuscript received October 16, 2016; revised November 29, 2016;

accepted December 7, 2016. Date of publication December 9, 2016; date

of current version April 7, 2017. This work was supported in part by

the National Natural Science Foundation of China (NSFC) under Grant

(91438117, 61671124, 61201129). The associate editor coordinating the

review of this letter and approving it for publication was K. Ota.

X. Zhang, R. Lin, and L. He are with the School of Communication

and Information Engineering and Center for Cyber Security, University

of Electronic Science and Technology of China, Chengdu 611731, China

(e-mail: xnzhang@uestc.edu.cn).

Z. Cheng is with Beijing Institute of Satellite Information Engineering,

Beijing 100191, China (e-mail: linuxdemo@126.com).

S. Yu is with the School of IT, Deakin University, Melbourne, VIC 3125,

Australia.

H. Luo is with the School of Electronic and Information Engineering,

Beijing Jiaotong University, Beijing 100044, China.

Digital Object Identiﬁer 10.1109/LCOMM.2016.2638430

However, in the case of restoration [4], recovery paths are not

allocated until a failure occurs. Thus, when a failure occurs,

additional signaling is required to establish restoration path.

Sharma et al. [4] proposed path restoration where the SDN

controller reacts to the failure notiﬁcation by deleting the ﬂow

entries of the disrupted ﬂows, computing restoration paths, and

installing new required ﬂow entries.

In the mechanism of SDN protection, ﬂow entries of a

backup path are installed in advance. When a link fails, the

data-plane switches automatically activate the backup path

without involving the SDN controller. Thus SDN protection

enables fast recovery from link failure. However, SDN pro-

tection possesses disadvantages as well. Firstly, the protection

technique itself is rigid. The working and backup paths are

correlated, as the latter is reassigned every time the working

path is changed. Secondly, ﬂow entries in a SDN switch

are stored in Ternary Content Addressable Memory (TCAM),

which is limited in size, and it can hold only a few thousands

of entries as it is expensive and power hungry. In SDN

protection, extra ﬂow entries for backup paths consume valu-

able TCAMs, which greatly increases network cost. In the

mechanism of SDN restoration, ﬂow entries of a restoration

path are installed only after the failure notiﬁcation. As a result,

the number of ﬂow entries is greatly reduced than that of

protection mechanism. But the recovery time from link failure

is larger than that of protection mechanism. The reason is that

the SDN controller calculates restoration paths and installs

new ﬂow entries for all disrupted ﬂows. Therefore, frequent

ﬂow operations are needed between the SDN controller and

switches, which leads to the increment of recovery time. From

the advantages and disadvantages of protection and restoration,

it is necessary to design a new resilient mechanism to balance

failure recover time efﬁciency and memory cost in SDN.

In this letter, we ﬁrst propose Local Fast Reroute (LFR)

algorithm in SDN. LFR is designed to provide local reroute

path for any possible link failure between adjacent nodes after

link failure occurring. All trafﬁc ﬂows affected by the failure

are aggregated to be a new “big” ﬂow with a new label. Then

the SDN controller only needs to calculate restoration path and

install ﬂow entries for the aggregation ﬂow. LFR enables fast

recovery by reducing the number of ﬂow operations between

the SDN controller and switches.

II. L

OCAL FAST REROUTE

The proposed Local Fast Reroute (LFR) algorithm requires

an immediate action of the SDN controller after notiﬁcation

of a link failure. The disrupted trafﬁc ﬂows are aggregated

with a new label. Then the SDN controller calculates a local

reroute path and installs new ﬂow entries for the aggregation

ﬂow. For brevity, we ﬁrstly summarize the notations that will

be used in Table I.

See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

下载后可阅读完整内容，剩余3页未读，立即下载

weixin_38744694

粉丝: 17
资源: 948

SDN中基于流量聚合的局部快速重路由算法

RFC8537 Fast Reroute Procedures for Co-routed Associated Bidirectional LSPs

MPLS Fast Reroute Network Operations Guide

Networkers2009：BRKIPM-3019 - Advanced Fast Convergence and IP FRR in IS-IS

BFD auto frr

ospf ip frp

FRR中有NSR功能吗

ospf ip frr

H3Cospf加快收敛

FRR和ECMP可以一起配置吗

如何设置DPDK支持

最新资源