Download Demonstrating the Optimal Placement of Virtualized Cellular

Demonstrating the Optimal Placement of Virtualized
Cellular Network Functions in Case of Large Crowd Events
Steffen Gebert, David Hock,
Marco Hoffmann,
Thomas Zinner,
Michael Jarschel,
Phuoc Tran-Gia
Ernst-Dieter Schmidt
University of Würzburg,
Germany
Ralf-Peter Braun
Deutsche Telekom T-Labs
Berlin, Germany
Nokia
Munich, Germany
Christian Banse
Andreas Köpsel
Fraunhofer AISEC
Garching, Germany
BISDN
Berlin, Germany
Categories and Subject Descriptors
To successfully deploy NFV-based EPC components, several challenges have to be addressed [2]. These include the
deployment, interconnection, and configuration of LTE components in the cloud. Therefore, entities that instantiate and
orchestrate the virtualized functions are required.
The presented demonstration illustrates a dynamic attachment of virtualized SGWs to an LTE network to support
additional users, cf., Figure 1. New SGWs are placed in the
network according to the varying traffic demands. This is
achieved by reprogramming SDN-enabled network elements
(NE+). Figure 1(a) shows the normal configuration of the
LTE network with the static SGW and the NE+ that can be
turned into virtual SGWs on demand. For better readability, non-reprogrammable network elements (NE) and base
stations omitted. Figure 1(b) shows the situation with dynamic function placement. Large crowd events, indicated
by stars of di↵erent sizes, lead to increased traffic demand.
By reprogramming several NE+, additional demands can be
handled. Depending on the number, size, and location of the
events, an adequate number of NE+ has to be selected.
C.2.3 [Computer Systems Organization]: ComputerCommunication Networks—Network Operations
Keywords
function placement, network functions virtualisation
1. INTRODUCTION
This demonstration highlights how Network Functions Virtualization (NFV) [1] can be used by a mobile network operator to dynamically provide required mobile core network
functions for a large “Mega” event like a football game or a
concert. Economic reasons may not justify the deployment
or continuous maintenance of expensive dedicated hardware
at the event site, which is necessary to cope with the extra
load temporarily generated by the visitors.
The Evolved Packet Core (EPC) in nowadays’ mobile LTE
networks consists of several, specialized components: first,
pure control plane elements like Mobility Management Entities (MME), which can be installed on virtualized IT hardware in the cloud. Second, gateways which are a mixture
of control and user plane. Such a component is the Serving Gateway (SGW), which switches GTP (GPRS Tunneling Protocol) tunnels in LTE networks. Usually, per ⇡ 10
mio. subscribers, 10 SGW devices are in use and statically
placed with respect to normal traffic demand. In case of
“Mega” events, however, the spatial distribution of the traffic demand changes, resulting in the need to dynamically
attach new SGWs to the LTE network for the duration of
the event. This needed flexibility and on-demand behavior
can be provided by virtualized network functions.
A
Home Area
Event Area
A B
Permission to make digital or hard copies of part or all of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for profit or commercial advantage, and that copies bear this notice and the full citation on the first page. Copyrights for third-party components of this work must be
honored. For all other uses, contact the owner/author(s). Copyright is held by the
author/owner(s).
SIGCOMM’14, August 17–22, 2014, Chicago, IL, USA.
ACM 978-1-4503-2836-4/14/08
http://dx.doi.org/10.1145/2619239.2631428 .
Event Area
B
(a) Normal configuration.
SGWs
Home Area
Reprogrammable NE+
(b) Adaptive configuration
using virtualized SGWs.
Virtual SGWs
Crowd Events
Figure 1: Dynamic function attachment to the LTE
core network.
359
In the following, the steps of the function placement are
described and the involved tools introduced. Then, a sketch
of the planned demonstration at SIGCOMM as well as the
demonstration setup is provided.
Home Ben
Ann
Data center
2. DYNAMIC FUNCTION PLACEMENT
SGW
Figure 2 indicates the di↵erent steps of the dynamic capacity addition and gives an overview of the involved components. First, the SGW placement has to be optimized.
In this demo, this step is performed by an extension of
the POCO tool [3, 4]. POCO allows the consideration of
trade-o↵s between di↵erent metrics, such as the number of
additional SGWs, the maximum latency to the gateways,
their processing capabilities, or load balancing on the SGWs.
As second and third step, the dynamically added applications are deployed, interconnected, monitored, and configured. This is done by NOKIA’s Cloud Application Manager
(CAM) [5] based on OpenStack. For network operations and
the triggering of CAM and POCO an NOKIA orchestration
tool called Network Utilization Controller (NUC) is used.
POCO
1. Optimize placement of new SGWs
2. Deploy SGW App and Controller
SGW
SGW
App Controller
4. Security check
3. Program virtual GW
Home Ben
CAM
Ann
Operator
Control Center
Video
call
Data center
SGW
Event
Video
call
SGW SGW SGW
App Controller
(b) Event situation. Ann is visiting a “Mega” event.
Figure 3: Demonstration scenario.
LTE emulator provided by Nokia, except the SGW. It is
an SDN-based gateway consisting of Ethernet/IP/GTP enabled NE, SDN Ethernet/IP/GTP controllers, and an SGW
application. The LTE emulator establishes a 3GPP compliant user and control plane communication in which the
SGW is integrated. Furthermore, base stations and user
equipments are part of the emulator. The components are
realized in software and are installed in the cloud environment of Deutsche Telekom T-Labs. The cloud is managed
by OpenStack and the CAM is used for application management. To show the video call, two tablets are connected via
a WiFi access point to the user equipment (endpoints) of the
emulator and WiFi traffic is translated into 3GPP compliant traffic. CAM, NUC and POCO have their own graphical
user interfaces that are integrated into a demo control GUI.
POCO
SDN+CAM
NE+
(a) Normal situation. Ann is at home.
For each
SGW
NUC
CAM
Operator
Control Center
Event
Operator
Control Center
Figure 2: Steps of dynamic capacity addition.
3. SCENARIO AND DEMO PRESENTATION
In the presented demonstration, a “Mega” event use case
is shown. Ann and Ben are living in a so-called home area
and are subscribers to an LTE network. Together with many
others, Ann is traveling to the event area, a stadium. Ben
stays in the home area. This is indicated by the locations of
the icons representing Ann and Ben in Figure 1. Figure 3
illustrates the considered scenario, including the normal situation Figure 3(a) and the event situation Figure 3(b).
The home area is a typical LTE network including base
stations, an evolved packet core, and data centers. OpenStack is used to manage the mobile operator’s cloud (data
centers). The mobile network operator is aware of the situation that subscribers in the event area will extensively share
pictures or videos and will make video calls resulting in a
higher network load. Therefore, the operator will activate
additional base stations that are installed in the event area,
but inactive during normal operations. Additionally, base
stations, e.g. mounted on vans, can be set up and connected
to the network. POCO selects optimal NE+ locations for
additional SGW, which will be activated using CAM by installing an SDN GTP controller and an SGW application in
a data center best located in relation to the network element.
After deployment, configuration, and activation of the
network element the event area is ready for the “Mega” event,
as illustrated by Figure 3(b). Then all visitors, including
Ann, enter the stadium, the “Mega” event starts, and Ann
places a video call to Ben who is still in the home area. After
the end of the “Mega” event, all visitors leave the event area
and the operator releases the deployed resources.
The demo emulates a real LTE network as closely as possible. All LTE network components are realized through an
4.
ACKNOWLEDGMENTS
This work has been performed in the framework of the
CELTIC EUREKA project SASER-SIEGFRIED (Project
ID CPP2011/2-5), and it is partly funded by the BMBF
(Project ID 16BP12308). The authors alone are responsible
for the content of the paper.
5.
REFERENCES
[1] “Network Functions Virtualisation - Introductory
White Paper,” 2012.
[2] A. Basta, W. Kellerer, H. J. Morper, and M. Ho↵mann,
“Applying NFV and SDN to LTE Mobile Core
Gateways; The Functions Placement Problem,” in
Accepted for ACM SIGCOMM 2014 Workshop All
Things Cellular, Chicago, USA, 2014.
[3] D. Hock, M. Hartmann, S. Gebert, M. Jarschel,
T. Zinner, and P. Tran-Gia, “Pareto-Optimal Resilient
Controller Placement in SDN-based Core Networks,” in
ITC, Shanghai, China, 2013.
[4] D. Hock, M. Hartmann, S. Gebert, T. Zinner, and
P. Tran-Gia, “POCO-PLC: Enabling Dynamic ParetoOptimal Resilient Controller Placement in SDN
Networks,” in INFOCOM, Toronto, Canada, 2014.
[5] NOKIA FutureWorks, “SDN in Mobile Networks,” in
Mobile World Congress, Barcelona, Spain, 2014.
360