TEAS Working Group A.Wang
Internet Draft China Telecom
Xiaohong Huang
BUPT
Caixia Kou
BUPT
Lu Huang
China Mobile
Penghui Mi
Tencent Company
Intended status: Information Track July 18, 2017
Expires: January 17, 2018
CCDR Scenario, Simulation and Suggestion
draft-wang-teas-ccdr-01.txt
Status of this Memo
This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79.
This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79. This document may not be modified,
and derivative works of it may not be created, and it may not be
published except as an Internet-Draft.
This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79. This document may not be modified,
and derivative works of it may not be created, except to publish it
as an RFC and to translate it into languages other than English.
it for publication as an RFC or to translate it into languages other
than English.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF), its areas, and its working groups. Note that
other groups may also distribute working documents as Internet-
Drafts.
Internet-Drafts are draft documents valid for a maximum of six
months and may be updated, replaced, or obsoleted by other documents
at any time. It is inappropriate to use Internet-Drafts as
reference material or to cite them other than as "work in progress."
The list of current Internet-Drafts can be accessed at
http://www.ietf.org/ietf/1id-abstracts.txt
<A.Wang> Expires January 17, 2018 [Page 1]
Internet-Draft CCDR Scenario, Simulation and Suggestion July 18, 2017
The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html
This Internet-Draft will expire on January 18, 2009.
Copyright Notice
Copyright (c) 2017 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with
respect to this document.
Abstract
This document describes the scenarios, simulation and suggestions
for the "Centrally Control Dynamic Routing (CCDR)" architecture,
which integrates the merit of traditional distributed protocols
(IGP/BGP), and the power of centrally control technologies (PCE/SDN)
to provide one feasible traffic engineering solution in various
complex scenarios for the service provider.
Traditional MPLS-TE solution is mainly used in static network
planning scenario and is difficult to meet the QoS assurance
requirements in real-time traffic network. With the emerge of SDN
concept and related technologies, it is possible to simplify the
complexity of distributed control protocol, utilize the global view
of network condition, give more efficient solution for traffic
engineering in various complex scenarios.
Table of Contents
1. Introduction ................................................ 3
2. Conventions used in this document............................ 4
3. CCDR Scenarios. ............................................. 4
3.1. Qos Assurance for Hybrid Cloud-based Application.........4
3.2. Increase link utilization based on tidal phenomena.......5
3.3. Traffic engineering for IDC/MAN asymmetric link..........6
3.4. Network temporal congestion elimination. ................6
4. CCDR Simulation. ............................................ 7
4.1. Topology Simulation..................................... 7
4.2. Traffic Matrix Simulation............................... 8
4.3. End-to-End Path Optimization............................ 8
4.4. Network temporal congestion elimination .................9
<A.Wang et.> Expires January 17, 2018 [Page 2]
Internet-Draft CCDR Scenario, Simulation and Suggestion July 18, 2017
5. CCDR Deployment Consideration............................... 11
6. Security Considerations..................................... 11
7. IANA Considerations ........................................ 11
8. Conclusions ................................................ 11
9. References ................................................. 11
9.1. Normative References................................... 11
9.2. Informative References................................. 12
10. Contributors: ............................................. 12
11. Acknowledgments ........................................... 13
1. Introduction
Internet network is composed mainly tens of thousands of routers that
run distributed protocol to exchange the reachability information
between them. The path for the destination network is mainly
calculated and controlled by the traditional IGP protocol. These
distributed protocols are robust enough to support the current
evolution of Internet but has some difficulties when the application
requires the end-to-end QoS performance, or the service provider
wants to maximize the links utilization within their network.
MPLS-TE technology is one perfect solution for the finely planned
network but it will put heavy burden on the router when we use it to
solve the dynamic QoS assurance requirements within real time traffic
network.
SR(Segment Routing) is another prominent solution that integrates
some merits of traditional distributed protocol and the advantages of
centrally control mode, but it requires the underlying network,
especially the provider edge router to do label push and pop action
in-depth, and need some complex solutions for co-exist with the Non-
SR network. Finally, it can only maneuver the end-to-end path for
MPLS and IPv6 traffic via different mechanism.
The advantage of MPLS is mainly for traffic isolation, such as the
L2/L3 VPN service deployments. With the emerge of cloud-based
services, especially the hybrid cloud communication services, the
customers requires mainly the end-to-end QoS assurance services
between their private infrastructure and the rented public servers.
Without the help of centrally control architecture, the service
provider almost can't make such SLA guarantees upon the real time
traffic situation.
This draft gives some scenarios that the centrally control dynamic
<A.Wang et.> Expires January 17, 2018 [Page 3]
Internet-Draft CCDR Scenario, Simulation and Suggestion July 18, 2017
routing (CCDR) architecture can easily solve, without adding more
extra burdening on the router. It also gives the PCE algorithm
results under the similar topology, traffic pattern and network size
to illustrate the applicability of CCDR architecture. Finally, it
gives some suggestions for the implementation and deployment of CCDR.
2. Conventions used in this document
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in RFC 2119 [RFC2119].
3. CCDR Scenarios.
The following sections describe some scenarios that the CCDR
architecture is suitable for deployment.
3.1. Qos Assurance for Hybrid Cloud-based Application.
With the emerge of cloud computing technologies, enterprises are
putting more and more services on the public oriented service
infrastructure, but keep still some core services within their
network. The bandwidth requirements between the private cloud and
the public cloud are occasionally and the background traffic between
these two sites varied from time to time. Enterprise cloud
applications just want to have the capabilities to invoke the
network to make the end-to-end QoS assurance on demand. Otherwise,
the traffic should be controlled by the default distributed protocol.
CCDR, which integrates the merits of distributed protocol and the
power of centrally control, is suitable for this scenario. The
possible solution architecture is illustrated below:
+------------------------+
| Cloud Based Application|
+------------------------+
|
+-----------+
| PCE |
+-----------+
|
|
//--------------\\
///// \\\\\
Private Cloud Site || Distributed |Public Cloud Site
| Control Network |
\\\\\ /////
\\--------------//
<A.Wang et.> Expires January 17, 2018 [Page 4]
Internet-Draft CCDR Scenario, Simulation and Suggestion July 18, 2017
Fig.1 Hybrid Cloud Communication Scenario
By default, the traffic path between the private cloud site and
public cloud site will be determined by the distributed control
network. When some applications require the end-to-end QoS assurance,
it can send these requirements to PCE, let PCE compute one e2e path
which is based on the underlying network topology and the real
traffic information, to accommodate the application's bandwidth
requirements. The proposed solution can refer the draft [draft-wang-
teas-pce-native-ip]. Section 4 describes the detail simulation
process and the results.
3.2. Increase link utilization based on tidal phenomena.
Currently, the network topology within MAN is generally in star
style as illustrated in Fig.2, with the different devices connect
different kind customer. The traffic pattern of these customers
demonstrates some tidal phenomena that the links between the CR/BRAS
and CR/SR will experience congestion in different periods because
the subscribers under BRAS often use the network at night and the
dedicated line users under SR often use the network during the
daytime. The uplink between BRAS/SR and CR must satisfy the maximum
traffic pattern between them and this causes the links utilization
always not efficient enough.
+--------+
| CR |
+----|---+
|
--------|--------|-------|
| | | |
+--|-+ +-|- +--|-+ +-|+
|BRAS| |SR| |BRAS| |SR|
+----+ +--+ +----+ +--+
Fig.2 STAR-style network topology within MAN
If we can consider link the BRAS/SR with local loop, and control the
MAN with the CCDR architecture, we can exploit the tidal phenomena
between BRAS/CR and SR/CR links, increase the efficiency of them.
+-------+
----- PCE |
| +-------+
+----|---+
| CR |
+----|---+
|
--------|--------|-------|
<A.Wang et.> Expires January 17, 2018 [Page 5]
Internet-Draft CCDR Scenario, Simulation and Suggestion July 18, 2017
| | | |
+--|-+ +-|- +--|-+ +-|+
|BRAS-----SR| |BRAS-----SR|
+----+ +--+ +----+ +--+
Fig.3 Increase the link utilization via CCDR
3.3. Traffic engineering for IDC/MAN asymmetric link
The operator's networks are often comprised by tens of different
domains, interconnected with each other, form very complex topology
that illustrated in Fig.4. Due to the traffic pattern to/from MAN
and IDC, the links between them are often in asymmetric style. It is
almost impossible to balance the utilization of these links via the
traditional distributed protocol, but this unbalance phenomenon can
be overcome via the CCDR architecture.
+---+ +---+
|MAN|-----------------IDC|
+-|-| | +-|-+
| ---------| |
------|BackBone|------
| ----|----| |
| | |
+-|-- | ----+
|IDC|----------------|MAN|
+---| |---+
Fig.4 TE within Complex Multi-Domain topology
3.4. Network temporal congestion elimination.
In more general situation, there are often temporal congestion
periods within part of the service provider's network. Such
congestion phenomena will appear repeatedly and if the service
provider has some methods to mitigate it, it will certainly increase
the satisfaction degree of their customer. CCDR is also suitable for
such scenario that the traditional distributed protocol will process
most of the traffic forwarding and the controller will schedule some
traffic out of the congestion links to lower the utilization of them.
Section 4 describes the simulation process and results about such
scenario.
<A.Wang et.> Expires January 17, 2018 [Page 6]
Internet-Draft CCDR Scenario, Simulation and Suggestion July 18, 2017
4. CCDR Simulation.
The following sections describe the topology, traffic matrix, end-
to-end path optimization and congestion elimination in CCDR
simulation.
4.1. Topology Simulation.
Technically, the topology involved nodes and links state information
is significantly helpful for traffic schedule.
The network topology mainly contains nodes and links information.
Nodes used in simulation have two types: core nodes and edge nodes.
The core nodes are fully linked to each other. The edge nodes are
connected with some of the core nodes. And edge nodes are not
connected with other edge nodes directly. Fig.5 is a topology
example of 4 core nodes and 5 edge nodes. In this simulation, 100
core nodes and 400 edge nodes are generated.
+----+
/|Edge|\
| +----+ |
| |
| |
+----+ +----+ +----+
|Edge|----|Core|-----|Core|---------+
+----+ +----+ +----+ |
/ | \ / | |
+----+ | \ / | |
|Edge| | X | |
+----+ | / \ | |
\ | / \ | |
+----+ +----+ +----+ |
|Edge|----|Core|-----|Core| |
+----+ +----+ +----+ |
| | |
| +------\ +----+
| ---|Edge|
+-----------------/ +----+
Fig.5 Topology of simulation
The total number of links is set to be more than 20000. The number
of links connecting one edge node to the set of core nodes is
randomly between 2 to 30. The bandwidth of all links is set to be
100Gbps. The metric of links between core nodes themselves is set to
be from 60 to 100, while metric of links between core nodes and edge
nodes is set to be from 1000 to 1060. The metric of links is used
<A.Wang et.> Expires January 17, 2018 [Page 7]
Internet-Draft CCDR Scenario, Simulation and Suggestion July 18, 2017
for selecting the shortest paths of all source-destination pairs.
Besides, each link has its congestion threshold. For the links
between core nodes, the threshold is set to be 0.8 which means when
its utilization is beyond 80% the link is overloaded. Otherwise, the
link is not congested. Similarly, the threshold of links between an
edge node and a core node is set to be 0.9.
4.2. Traffic Matrix Simulation.
The end-to-end traffic of the network is regard as a n*n matrix
where n stands for the number of forwarding devices in the network.
Each (i,j) component of traffic matrix denotes the bandwidth of the
flow from i-th node to j-th node. The traffic matrix is generated
based on the link capacity of topology. It can result in many kinds
of situations, such as congestion, mild congestion and non-
congestion.
In this simulation, the traffic matrix is 500*500. The components of
traffic matrix are generated from 10Mbps to 7Gbps randomly. About 20%
links are overloaded when the Open Shortest Path First (OSPF)
protocol is used in the network.
This traffic matrix is used in following sections. In section 4.3,
it is used as the background traffic which can't be scheduled. In
section 4.4, it is re-routed based on load-balance.
4.3. End-to-End Path Optimization
Based on the current state of the network, such as the traffic
matrix in the network, network topology and network utilization,
Quality of Service (QoS) and so on, the end-to-end path optimization
is to find the best end-to-end path which is the lowest in metric
value and each link of the path is far below link's threshold. The
algorithm is a novel idea combining the shortest path algorithm with
penalty theory of classical optimization and graph theory.
Given background traffic matrix which is unscheduled, when a set of
new flows comes into the network the end-to-end path optimization
finds the optimal paths for them. The selected paths bring the least
congestion degree to the network.
The simulation is tested with 1000 flows in 6 periods. The size of
flows is from 10Mbps to 10Gbps. In each period, 100, 200, 100, 250,
150 and 200 flows are arrived respectively. The link utilization
increment(UI) degree relative to the congestion threshold when the
new flows are added into the network is shown in Fig.6. The first
graph in Fig.6 is the UI with OSPF and the second graph is the UI
with end-to-end path optimization. The average UI of graph one is
more than 30%. After path optimization as shown in graph, the
<A.Wang et.> Expires January 17, 2018 [Page 8]
Internet-Draft CCDR Scenario, Simulation and Suggestion July 18, 2017
average UI is less than 5%. In a conclude, the results show that the
end-to-end path optimization has an eye-catching decreasing in UI
degree relative to the path chosen based on OSPF.
+-----------------------------------------------------------+
| * * * *|
60| * * * * * *|
|* * ** * * * * * ** * * * * **|
|* * ** * * ** *** ** * * ** * * * ** * * *** **|
|* * * ** * ** ** *** *** ** **** ** *** **** ** *** **|
40|* * * ***** ** *** *** *** ** **** ** *** ***** ****** **|
UI(%) |* * ******* ** *** *** ******* **** ** *** ***** *********|
|*** ******* ** **** *********** *********** ***************|
|******************* *********** *********** ***************|
20|******************* ***************************************|
|******************* ***************************************|
|***********************************************************|
|***********************************************************|
0+-----------------------------------------------------------+
0 100 200 300 400 500 600 700 800 900 1000
+-----------------------------------------------------------+
| |
60| |
| |
| |
| |
40| |
UI(%) | |
| |
| |
20| |
| *|
| * *|
| * * * * * ** * *|
0+-----------------------------------------------------------+
0 100 200 300 400 500 600 700 800 900 1000
Flow Number
Fig.6 Simulation result with congestion elimination
4.4. Network temporal congestion elimination
In general situation, there are often temporal congestion periods
within part of the service provider's network. The network temporal
congestion elimination is proposed which reroutes traffic from the
congested paths to un-congested ones. The load-balance is achieved
<A.Wang et.> Expires January 17, 2018 [Page 9]
Internet-Draft CCDR Scenario, Simulation and Suggestion July 18, 2017
after congestion elimination. And the cost of reroute traffic is
also taken into consideration.
Different degree of network congestion is simulated. About 20% links
are congested with slightly or badly degree using the OSPF protocol.
The congestion degree (CD) is defined as the link utilization beyond
its threshold. For example, if the utilization of links is 90%, and
its threshold is 80%, then its CD is 10%.
The congestion elimination performance is shown in Fig.7. The first
graph is the congestion degree before the process of congestion
elimination. The average CD of all congested links is more than 10%.
The second graph shown in Fig.7 is the congestion degree after
congestion elimination process. It shows only 12 links among totally
2000 links exceed the threshold, and all the congestion degree is
less than 3%. Thus, after schedule of the traffic in congestion
paths, the degree of network congestion is greatly eliminated and
the network utilization is indeed in balance.
Before congestion elimination
+-----------------------------------------------------------+
| * ** * ** ** *|
20| * * **** * ** ** *|
|* * ** * ** ** **** * ***** *********|
|* * * * * **** ****** * ** *** **********************|
15|* * * ** * ** **** ********* *****************************|
|* * ****** ******* ********* *****************************|
CD(%) |* ********* ******* ***************************************|
10|* ********* ***********************************************|
|*********** ***********************************************|
|***********************************************************|
5|***********************************************************|
|***********************************************************|
|***********************************************************|
0+-----------------------------------------------------------+
0 0.5 1 1.5 2
After congestion elimination
+-----------------------------------------------------------+
| |
20| |
| |
| |
15| |
| |
CD(%) | |
10| |
| |
| |
<A.Wang et.> Expires January 17, 2018 [Page 10]
Internet-Draft CCDR Scenario, Simulation and Suggestion July 18, 2017
5 | |
| |
| * ** * * * ** * ** * |
0 +-----------------------------------------------------------+
0 0.5 1 1.5 2
Link Number(*10000)
Fig.7 Simulation result with congestion elimination
5. CCDR Deployment Consideration.
With the above scenarios and simulation results, we can know it is
necessary to find one general solution to cope with various complex
situations and it is possible to accomplish the most complex optimal
path computation function in centrally manner based on the underlay
network topology and the real time traffic.
[draft-wang-teas-native-ip] gives one basic solution for above
scenario, such thought can be extended to cover requirements that
are more concretes.
6. Security Considerations
TBD
7. IANA Considerations
TBD
8. Conclusions
TBD
9. References
9.1. Normative References
[RFC4655] Farrel, A., Vasseur, J.-P., and J. Ash, "A Path
Computation Element (PCE)-Based Architecture", RFC
4655, August 2006,<http://www.rfc-editor.org/info/rfc4655>.
[RFC5440]Vasseur, JP., Ed., and JL. Le Roux, Ed., "Path
Computation Element (PCE) Communication Protocol
(PCEP)", RFC 5440, March 2009,
<A.Wang et.> Expires January 17, 2018 [Page 11]
Internet-Draft CCDR Scenario, Simulation and Suggestion July 18, 2017
<http://www.rfc-editor.org/info/rfc5440>.
9.2. Informative References
[I-D.draft-ietf-teas-pce-control-function]
A. Farrel, Q.Zhao et al. "An Architecture for use of PCE and PCEP in
a Network with Central Control"
https://datatracker.ietf.org/doc/draft-ietf-teas-pce-central-
control/ September, 2016
[I-D. draft-ietf-teas-pcecc-use-cases]
Quintin Zhao, Robin Li, Boris Khasanov et al. "The Use Cases for
Using PCE as the Central Controller(PCECC) of LSPs
https://tools.ietf.org/html/draft-ietf-teas-pcecc-use-cases-00
March,2017
[I-D. draft-wang-teas-pce-native-ip]
A.Wang, Quintin Zhao, Boris Khasanov, Penghui Mi,Raghavendra Mallya,
Shaofu Peng "PCE in Native IP Network"
https://tools.ietf.org/html/draft-wang-teas-pce-native-ip-03 March
13, 2017
[I-D. draft-wang-pcep-extension for native IP]
Aijun Wang, Boris Khasanov et al. "PCEP Extension for Native IP
Network" https://datatracker.ietf.org/doc/draft-wang-pce-extension-
native-ip/
10. Contributors:
Xiaoyan Wei
<A.Wang et.> Expires January 17, 2018 [Page 12]
Internet-Draft CCDR Scenario, Simulation and Suggestion July 18, 2017
China Telecom Shanghai Company
weixiaoyan@189.cn
Qiong Sun
sunqiong.bri@chinatelecom.cn
Tingting Yuan
Beijing University of Posts and Telecommunications
yuantingting@bupt.edu.cn
Dingyuan Hu
Beijing University of Posts and Telecommunications
hdy@bupt.edu.cn
11. Acknowledgments
TBD
Authors' Addresses
Aijun Wang
China Telecom
Beiqijia Town, Changping District
Beijing,China
Email: wangaj.bri@chinatelecom.cn
Xiaohong Huang
Beijing University of Posts and Telecommunications
No.10 Xitucheng Road, Haidian District
Beijing,China
EMail: huangxh@bupt.edu.cn
<A.Wang et.> Expires January 17, 2018 [Page 13]
Internet-Draft CCDR Scenario, Simulation and Suggestion July 18, 2017
Caixia Kou
Beijing University of Posts and Telecommunications
No.10 Xitucheng Road, Haidian District
Beijing,China
koucx@lsec.cc.ac.cn
Lu Huang
China Mobile
32 Xuanwumen West Ave, Xicheng District
Beijing 100053
China
Email: hlisname@yahoo.com
Penghui Mi
Tencent
Tencent Building, Kejizhongyi Avenue,
Hi-techPark, Nanshan District,Shenzhen 518057, P.R.China
Email kevinmi@tencent.com
<A.Wang et.> Expires January 17, 2018 [Page 14]