Real-World IPTV Network Measurements Georgios Baltoglou Eirini Karapistoli Periklis Chatzimisios School of Technology and Health Department of Electronics Department of Informatics KTH, Stockholm, Sweden Alexander TEI of Thessaloniki, Greece Alexander TEI of Thessaloniki, Greece [email protected] [email protected] [email protected] Abstract—Internet Protocol Television or IPTV is defined as while an over-provisioned network might be expensive to an alternate method of distributing television content over IP by deploy or have extravagant leasing costs. Finally, throughout a telecom carrier or an Internet service provider (ISP) and it measurement data the optimal time for upgrade and the extent is increasingly gaining market share in modern communication networks. IPTV poses specific requirements on the network in- of the upgrade can be decided. frastructure and at the same time it requires that several network The relation between the service providers or telcos, and the performance characteristics are maintained under defined levels end users can be summarized by what is called Service-Layer in order for the end user to have an assured viewing experience. Agreement or SLA. The SLA is a form of contract containing In this paper, our main motivation is to investigate how a real the responsibilities of both sides and the definition of the level life network performs in terms of distributing this inelastic and high-bandwidth type of service utilizing traffic measurements. of service that has to be provided by the telcos and experienced Furthermore, our goal is to verify and analyze whether the by the subscribed users. Again, measurements are of use.In studied network is suited for multicast IPTV traffic. order for the users to be able to observe the compliance of the operator to the agreed level of provided service. In contrast I. INTRODUCTION to telcos, users are rarely interested in quantitative network The latest entry into the television programming industry are performance metrics (i.e. packet loss rate, packet delay, delay telecommunication companies or telcos - which have started variation, etc.). What is of interest to them is the application providing television content over the internet commonly re- performance, a set of qualitative measurements that perplex the ferred to as IPTV. IPTV is a service with great expectations. QoS measured by the operators with the overall experience of It thrives to replace the former methods of TV broadcasting the end user. Modern traffic measurement techniques directly such as over the air, cable or satellite TV distribution, making provide QoE results, allowing the users to evaluate if their use of the packet-switched Internet infrastructure. In order current level of service is in par with their expected level of to succeed, it also provides a variety of new services such experience. as Video-on-Demand (VoD) and VCR-like functionality for This paper is organized as follows. In Section II, we the distributed TV media, together with the ability to co-exist discuss the work that has already been accomplished and with high rate data surfing and VoIP traffic in what is called the contribution of our paper. In Section III we describe a triple-play service. Yet, the highly maladaptive to network the architecture of the studied real-world, commercial IPTV deficiencies IPTV streams pose great demands in all separate network. In Sections IV and V we provide information on layers of the network architecture. why active measurements are ideally suited for IPTV traffic In modern network engineering problems dealing with along with a description of the equipment used for holding inelastic traffic, such as IPTV, the efficient use of network the measurement sessions, as well as the metrics that we resources that satisfy the expected levels of Quality of Service specifically focus on to evaluate the network’s performance in (QoS) is a major issue. At the same time, Quality of Experi- IPTV deployment. Finally, Section VI illustrates the obtained ence (QoE) or how the QoS metrics are perceived by the end real-time measurement results, accompanied by evaluation users is also of vital importance. In order to respond to that reports. The conclusion of the paper is found in Section VII. and to characterize the performance of an IP network, traffic measurements are utilized, which play very important roles for II. IPTV MEASUREMENTS AND CONTRIBUTION both the ISPs, network and telecommunication operators, as well as for the end users and even the measurement equipment Obtaining traffic measurements, especially for IPTV, inflicts vendors. certain problems that need to be researched: From the telcos’ perspective, network measurements pro- • Different types of network measurements exist, namely vide the means for network monitoring. As different important active and passive, each having its own advantages and metrics are gathered, such as traffic load and packet delay the disadvantages. As IPTV multicast traffic is peculiar and operators can identify the current state of the network and even demanding, not all traffic measurements are suited for predict eminent failures prior to happening. At the same time assessing the IPTV’s network performance. measurements are useful for network dimensioning. An under- • Real life networks resemble great heterogeneity. In terms provisioned network would result in dissatisfied customers of QoS implementation, multiple architectures exist such 978-1-4577-0681-3/11/$26.00 ©2011 IEEE 830 4$('%( . / . %0.1 (.23% %0.1 $ %0.1 ! !" # $ " +, $%$&'()$& " *$' & & (4 56-)7 ,)7 ,-- ,)7 *7 7 O2P ,-- 7 O$P ,7 Fig. 1. Studied Network Map as ATM, MPLS, IPv4, IPv6, and others. As such, ob- III. STUDIED NETWORK taining measurements of only a specific network layer In this section the exact network architecture of the real-life provides little with regard to the end-user QoS/QoE. At commercial IPTV network, on which our measurements were the same time, end-to-end measurements including totally conducted. A full network map was designed for this reason, unknown portions of a network provides poor information depicted in Fig. 1. The IPTV streams in the studied network for the subsequent analysis of the obtained data. are originating from Stockholm, Oslo and Copenhagen. The • While substantial work on traffic measurements and net- transportation and interconnection of these cities is provided work performance concerning IPTV has been made ([1] by the network provider Norrsken. Norrsken uses optical fibers and [2]), they focus on P2P traffic and passive mea- to connect several cities within Sweden and in Scandinavia surements, or they are usually produced via simulation in general, in an SDH ring topology. They use STM-16 software ([3] and [4]), mathematical modeling [5], or transmission frames, yielding a bitrate of 2.4 Gbps. The IPTV studying proof of concept lab networks, which have a streams transported to the city Borl¨ange are by the service more predictable behavior [6]. providers Canal-Digital and FastTV. Canal-Digital is transmit- • Studying only the performance of a high-level QoS ting multicast IPTV streams with 700 Mbps bandwidth, while priority class, as given in IPTV multicast traffic, is usually FastTV uses 500 Mbps. not enough for the end-user expectations. As IPTV is marketed in triple play bundles, including VoIP and data A. Metro / Core traffic, successive studies on these lower priority classes The Metro/Core layer is a Gigabit Ethernet network over are important, since they might be affected by the high optical fibers interconnecting cities Borl¨ange (where the IPTV priority class. stream is firstly received) and Falun. The link has a bandwidth of 10 Gbps and on its edges lie Cisco 7600 Series routers. From the above analysis, it is evident that no compre- These routers were specifically chosen when the network hensive study that we know of, has so far presented active was designed because they deploy high-performance Carrier measurements of a real-world commercial IPTV network. As Ethernet, IP/MPLS features and Enterprise WAN aggregation such, this is our work’s main contribution. Furthermore, with and core routing over 10 Gbps interfaces. Additionally, they the obtained results of our measurement scenarios and their support interface processors capable of controling voice, video, subsequent analysis we intend to: and data, offering a rich set of QoS features as specifically needed. Moreover, the Metro/Core routers have been config- ured with MPLS creating a QoS-guaranteed network. a provide tangible information and present the reasons why a network architecture as the one studied, verifies as being B. Distribution an exemplary IPTV distribution infrastructure. The Distribution network is composed of 1 Gbps Ethernet b be able to pinpoint the parts of the network that are links carried over optical fiber. The Distribution network sources of problems for IPTV delivery. situated in Falun consists of five redundantly interconnected c use them as a reference point for evaluating other net- Cisco 3750 ME switches featuring hierarchical QoS, traffic works and assessing the distributed IPTV service quality. shaping, intelligent 802.1Q tunneling, VLAN mapping and 978-1-4577-0681-3/11/$26.00 ©2011 IEEE 831 MPLS. Spanning Tree Protocol is further implemented to V. M ETRICS OF INTEREST prevent switching loops. All provided services are divided Due to the fundamental design of IP networks as a best- into different VLANs and each VLAN supports different QoS effort delivery platform independent of application require- classes. The highest priority class is given to the IPTV VLAN ments, packets and data are susceptible to loss. For the inelas- followed by VoIP, Management and finally the Internet VLAN. tic nature of the IPTV service, packet losses are a major issue. As mentioned, IPTV deployments commonly transport packets C. Access via UDP, providing no protection to packet loss. Packet loss of the audio stream can be exhibited as dropouts, squeaking On the Falun Access network the nodes are connected noises, variations in sound volume, and skipping.