PPSP Y. Zhang Internet Draft China Mobile Intended status: Informational N.Zong HuaweiTech G.Camarillo Ericsson J.seng PPlive R.Yang Yale University Expires: July 16, 2011 January 16, 2011 Problem Statement of P2P Streaming Protocol (PPSP) draft-ietf-ppsp-problem-statement-01 Abstract P2P streaming systems show more and more popularity in current Internet with proprietary protocols. This document identifies problems of the proprietary protocols, proposes standard signaling protocols called PPSP and discusses the scope and use case of PPSP. Status of this Memo This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet- Drafts is at http://datatracker.ietf.org/drafts/current/. 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." This Internet-Draft will expire on July 16, 2011. Zhang Expires July 16, 2011 [Page 1] Internet-Draft Problem Statement of PPSP January 2011 Copyright Notice Copyright (c) 2010 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. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License. Zhang Expires July 16, 2011 [Page 2] Internet-Draft Problem Statement of PPSP January 2011 Table of Contents 1. Introduction ................................................. 4 2. Terminology and concepts ..................................... 6 3. Problem statement ............................................ 7 3.1. ISP's difficulties in deploying P2P caches .............. 7 3.2. Difficulties in building open streaming delivery infrastructure ............................................... 7 3.3. Difficulties in mobile and wireless environment...........8 3.4. Terminal physical resource starvation ................... 9 4. PPSP:Standard peer to peer streaming protocols .............. 10 5. Use cases of PPSP ........................................... 13 5.1. Worldwide provision of open P2P live streaming services. 13 5.2. CDN supporting P2P streaming ........................... 13 5.3. PPSP supporting cross-screen streaming in heterogeneous environment ................................................. 14 5.4. Supporting P2P streaming in cellular mobile network..... 15 5.5. Cache service supporting P2P streaming ................. 16 6. Security Considerations ..................................... 17 6.1. Tracker Protocol ....................................... 17 6.2. Peer Protocol .......................................... 17 7. Acknowledgements ............................................ 18 8. References .................................................. 19 Zhang Expires July 16, 2011 [Page 3] Internet-Draft Problem Statement of PPSP January 2011 1. Introduction Streaming traffic is among the fastest growing traffic on the Internet. As Cisco Visual Network Traffic index measured, video streaming already generates the largest volume of Internet traffic in the year of 2010, and the percentage is expected to rise to as high as 91% of the total Internet traffic by 2014[Cisco]. There are two basic architectures for delivering streaming traffic on the global Internet: the client-server paradigm and the peer to peer (P2P) paradigm [Survey].The basic advantage of the P2P paradigm is its scalability and fault tolerance against failures of centralized infrastructures. As an example, PPLive [PPLive], one of the largest P2P streaming vendors, is able to distribute large-scale, live streaming programs such as the CCTV Spring Festival Gala to more than 3 million users with only a handful of servers. It can also deliver VoD streaming to a scale of some hundred of thousands simultaneous users using the same structure and similar protocols[VoD]. The effect of P2P technologies is also well demonstrated in delivering real and VoD streaming effectively in current practice like CNN [CNN] PPstream [PPstream],UUSee [UUSee]and CNTV[CNTV]. The latest release of Adobe Flash, a major platform of streaming distribution in the Internet, has also introduced Cirrus [Cirrus], a peer assisted data exchange mode. Almost all these systems use their proprietary signaling protocols and standard data delivery protocols (like UDP and TCP). What's more, along with the new players like CDN vendors (e.g.,Akamai [Akamai], ChinaCache[ChinaCache]) and ISPs(e.g., ComCast [ComCast])joining in the P2P streaming delivery game, the P2P streaming ecosystem is becoming more complex with participants from the source, infrastructure side, edge delivery even to the terminals. Given the above increasing integration of P2P streaming into the global content delivery infrastructure, the lacking of an open, standard P2P streaming signaling protocol suite becomes a major missing component in the Internet protocol stack. Multiple, similar but proprietary signaling protocols result in repetitious development efforts for new systems, and the lock-in effects lead to substantial difficulties in the integration. For example, in the enhancement of existing caches and CDN systems to support P2P streaming, the open protocols will dynamically reduce the complexity of the interaction with different P2P streaming applications. In this document we propose an open P2P streaming protocol named PPSP, to standardize signaling operations on two important components, peer and tracker for information exchange. The problems of proprietary Zhang Expires July 16, 2011 [Page 4] Internet-Draft Problem Statement of PPSP January 2011 signaling protocols and benefit of PPSP are explained further in section 3. PPSP will serve as an enabling technology, building on the development experiences of existing P2P streaming systems. Its design will allow it to integrate with IETF protocols on distributed resource location, traffic localization, and streaming control and data transfer mechanisms for building a complete streaming system or a streaming delivery infrastructure. Zhang Expires July 16, 2011 [Page 5] Internet-Draft Problem Statement of PPSP January 2011 2. Terminology and concepts Chunk: A chunk is a basic unit of partitioned streaming. Peers may use a chunk as a unit of storage, advertisement and exchange among peers [VoD]. Note that a streaming system may use different units for advertisement and data exchange, using chunks during data exchange, and a larger unit such as a set of chunks during advertisement. Content Distribution Network (CDN): A CDN node refers to a network entity that is deployed in the network (e.g., at the network edge or data centers) to store content provided by the original servers, and serves content to the clients located nearby topologically. Live streaming: It refers to a scenario where all clients receive streaming content for the same ongoing event. It is desired that the lags between the play points of the clients and that of the streaming source be small. P2P cache: A P2P cache refers to a network entity that caches P2P traffic in the network, and either transparently or explicitly as a peer distributes content to other peers. Peer: A peer refers to a participant in a P2P streaming system that not only receives streaming content, but also stores and uploads streaming content to other participants. PPSP: The abbreviation of P2P streaming protocols. PPSP refer to the key signaling protocols among various P2P streaming system components, including the tracker and the peer. Swarm: A swarm refers to a group of clients (i.e., peers) who exchang data to distribute the same content (e.g. video/audio program, digital file, etc) at a given time. Tracker: A tracker refers to a directory server which maintains a list of peers storing chunks for a specific channel or streaming file, and answers queries from peers for peer lists. The tracker is a logic component which can be centralized or distributed. Video-on-demand (VoD): It refers to a scenario where different clients may watch different parts of the same recorded media content during a past event. Zhang Expires July 16, 2011 [Page 6] Internet-Draft Problem Statement of PPSP January 2011 3. Problem statement Let's take a look what problems proprietary signaling brings out for P2P streaming applications. 3.1. ISP's difficulties in deploying P2P caches Facing with many P2P streaming applications, ISPs are witnessing a big traffic tension on their backbone and inter-networking ports.P2P cache is used to reduce the traffic by dynamically storing the frequently accessed streaming content (maybe in chunk or in file granularity). It's often deployed in the edge or inter-networking places of the network. However, the cache nodes need to execute DPI (deep packet inspection) for identifying different P2P streaming systems. Multiple ever changing proprietary P2P streaming protocols require the P2P cache updating its matching library constantly which increases the operator's cost dramatically. With PPSP, P2P cache can detect P2P streaming applications much easier. This reduces the ISP workload to a large extent. 3.2. Difficulties in building open streaming delivery infrastructure The future Internet is content-centric [CCN][DONA]because the vast majority of current Internet usage (a "high 90% level of traffic"[Van]) consists of data. Most of the content-centric works are seeking for building an open global content delivery infrastructure. In content-centric networks there is no doubt that P2P streaming data is going to account for a large portion. If current multiple proprietary protocols continue to work, there will exist lots of specific and independent systems to deliver vast streaming content. This brings more burdens for identifying and sharing the same contents, increases the storage, forwarding and maintenance cost in the intermediate nodes for repeated content and for the same content, many on-the-way data spread over the Internet. All these will definitely increase the cost of streaming distribution and causes possible congestion in the network. In an open P2P streaming delivery infrastructure, It must involve different participants in the distribution chain ranging from the source (or current P2P streaming vendors) , infrastructure (e.g, CDN) and delivery side(caches or terminal peers). Let's first consider a simplest open P2P streaming environment where different source vendors spread in different regions cooperatively deliver a broadcasting event. Suppose PPLive cooperatively broadcasts live Chinese spring festival gala for American Chinese with Pando Zhang Expires July 16, 2011 [Page 7] Internet-Draft Problem Statement of PPSP January 2011 networks. At a first sight, this is reasonable because PPlive has relatively few users in USA and Utilizing Pando's peer resources can help to realize efficient streaming delivery. However, different messages and interaction semantics between the two systems is a big challenge for the cooperative delivery. Consider a more complex case where source vendors cooperate with CDN providers. Such integration is already practiced by UUSee[UUSee], RayV[RayV] and Forcetech[Forcetech]. The effect of the involvement of infrastructure devices such as CDN nodes has been verified to improve the total performance of P2P streaming (e.g., with lower latency) by providing more stable "super peers" and reduce traffic in ISP network [CDN+P2P] [RFC 5693].However, there are substantial obstacles in deploying infrastructure nodes supporting proprietary P2P streaming protocols [HTPT]. Unlike the Web who has already equipped with the standard HTTP protocol and all kinds of the infrastructure devices support the protocol, in order to support P2P streaming, the infrastructure devices need to understand and keep updated with various protocols to take part in the delivery. Similar to the cache case, this introduces complexity and deployment cost for the infrastructure devices. With PPSP, CDN nodes can be designed to inter-operate with only the standard protocols, reducing the case by case negotiation between the source providers and multiple CDN providers. 3.3. Difficulties in mobile and wireless environment Mobility and wireless are becoming increasingly important features to support in future Internet deployments [GENI], [FIND]. Currently there are more and more mobile and wireless Internet users. It is predicted that by the end of 2012, the number of mobile Internet users will surpass that of fixed Internet users in China [Statistics]. Along with this trend, mobile streaming is becoming a key offering [MobileTV]. In Korea the number of mobile TV subscriber has reached seventeen millions, accounting for one third of the mobile subscribers. During the 2008 Beijing Olympic Games, more than one million users enjoyed mobile TV service. Although mobile handsets are more eligible to be peers with much better bandwidth and higher CPU frequency, larger storage and memory than several years ago, mobile and wireless peers may face bigger challenges for supporting P2P streaming with unsteady and relatively lower (esp. in uplink) network connections, less steady power than PCs. And what are worse, current protocols are designed mainly for the fixed Internet and do not address these challenges. Zhang Expires July 16, 2011 [Page 8] Internet-Draft Problem Statement of PPSP January 2011 In what kinds of network dynamical conditions, what kinds of mobile nodes can act as peers should be investigated to reflect in the standard protocols from the beginning of the design. 3.4. Terminal physical resource starvation Private protocols may require a terminal to install multiple different software at the same time. For example a user installs CBox for CCTV programs [CNTV], and PPLive for Japanese and Korean movies [PPLive]. However it may be difficult to install multiple clients in one resource constraint peer like mobile handsets or Pads. The limited CPU, storage and memory often limit the total number of installed applications as well as concurrent threads and processes. Note that for many client software, even it's not used by the users right now, the daemon may be invoked to facilitate other peers for free data delivery assistance. Standard protocols reduce the complexity of coexisting multiple closed systems and create possibilities to use one client software accommodating different systems. Zhang Expires July 16, 2011 [Page 9] Internet-Draft Problem Statement of PPSP January 2011 4. PPSP:Standard peer to peer streaming protocols The objective of this working group is to design PPSP, unified peer to peer streaming protocols to address the problems discussed in the preceding section. There are basically two kinds of P2P streaming systems, pull-based and push-based. In pull-based P2P streaming systems, a centralized tracker or distributed trackers maintains information about which peers are in which swarms and answers the peers' query on such information with a peer-list. After receiving the message, the peer can connect with the candidates in a swarm, exchange its content availability in its memory or storage (depending on it's real-time or VoD streaming) with other peers and then retrieve for wanted streaming data. The swarm is a mesh topology. Most of the current practices are belonging to this genre. The advantages of pull-based mode are its robustness to the peer churn and acceptable latency for a smooth play. On the other hand, in push-based P2P streaming systems, there is a head node maintaining the topology e.g., a tree. The peers in this topology share the same interest on content. The signaling and data distribution are both based on this topology. For one program or video file, the peer queries the head node for its location to join and the head node replies with a peer-list(maybe with recommended connection order). After receiving this peer-list, the peer can connect with the candidates for being a node in certain place of the topology and receive the data along this topology without the need of exchanging content availability with its siblings. In this sense the head node is acting as the tracker. The push mode has the advantages of lower latency but the topology is fragile to the peer churn. Few practical systems use this mode. A more practical mode is a hybrid pull-push mode where the peers exchange content availability with its siblings for retrieving unfounded data. As discussed above, in essence, there are two important entities in P2P streaming, i.e., trackers and peers. PPSP is targeted to standardize the signaling protocols in tracker-based architectures for supporting both live and VoD streaming. In live streaming, all peers are interested in the media coming from an ongoing event, which means that all peers share nearly the same streaming content at a given point of time. In live streaming, some Zhang Expires July 16, 2011 [Page 10] Internet-Draft Problem Statement of PPSP January 2011 peers may store the live media for further distribution, which is known as TSTV (time-shift TV), where the stored media are separated into chunks and distributed in a VoD-like manner. In VoD, different peers watch different parts of the recorded media content during a past event. In this case, each peer keeps asking other peers which media chunks are stored in which peers, and then gets the required media from certain/selected peers. In detail, PPSP designs a protocol for signaling between trackers and peers (the PPSP "tracker protocol") and a signaling protocol for communication among the peers (the PPSP "peer protocol") as shown in Figure 1. The two protocols enable peers to receive streaming data within the time constraints required by specific content items. The tracker protocol handles the initial and periodic exchange of meta information between trackers and peers, such as peer-list and content information. The peer protocol controls the advertising and exchange of media data between the peers. Note that in the pull mode and hybrid pull-push mode, both tracker protocol and peer protocol can be used; while in the push mode, only tracker protocol is used. What's more, existing protocols should be investigated and evaluated for being reused or extended as the protocols among peers, e.g., HTTP. Considering that there can be a large number of peers, the protocol should consider some lightweight (possibly binary) encoding. Zhang Expires July 16, 2011 [Page 11] Internet-Draft Problem Statement of PPSP January 2011 +------------------------------------------------+ | | | +--------------------------------+ | | | Tracker(Head Node) | | | +--------------------------------+ | | | ^ ^ | |Tracker | | Tracker |Tracker | |Protocol| | Procotol |Protocol | | | | | | | V | | | | +---------+ Peer +---------+ | | | Peer |<----------->| Peer | | | +---------+ Protocol +---------+ | | | ^ | | | |Peer | | | |Protocol | | V | | | +---------------+ | | | Peer | | | +---------------+ | | | | | +------------------------------------------------+ Figure 1 PPSP System Architecture Zhang Expires July 16, 2011 [Page 12] Internet-Draft Problem Statement of PPSP January 2011 5. Use cases of PPSP 5.1. Worldwide provision of open P2P live streaming services The cooperative vendors can easily expand the broadcasting scale with PPSP. In figure 2 the interactions between vendor A's tracker and vendor B and vendor C's super-nodes (SN in short) can be normalized using tracker protocol; and peer protocol can be used among SNs/peers spread in different vendors. +-------------------------------------------------------------------+ | | | +------------------+ | | +------------>| A's Tracker |<----------+ | | | +------------------+ | | | Tracker| ^ ^ | | | Protocol| Tracker| |Tracker |Tracker | | | Protocol| |Protocol |Protocol | | | | | | | | | | | | | | v v v v | | +------+ Peer +------+ +------+ +------+ | | | B's |<------->| B's | | C's | | C's | | | | SN1 |Protocol | SN2 | | SN1 | | SN2 | | | +------+ +------+ +------+ +------+ | | ^ ^ ^ ^ | | | | | | | | | | Peer Protocol Peer Protocol| | | | Peer | +-------------+ +--------------+ |Peer | | Procotol| | | |protocol| | | | | | | | | | | | | | | | | | | | v v v v | | +------+ Peer +------+ +---------+ Peer +---------+ | | | A's |<------> | B's | |A's |<------> |C's | | | | User1|Protocol | User2| | User1 |Protocol | User2 | | | +------+ +------+ +---------+ +---------+ | | | +-------------------------------------------------------------------+ Figure 2 Cooperative Vendors Interactions 5.2. CDN supporting P2P streaming This scenario is similar to use case 1 but it involves more kinds of participants: besides P2P streaming vendors, non-P2P streaming vendors and CDN providers are also involved in the delivery. Zhang Expires July 16, 2011 [Page 13] Internet-Draft Problem Statement of PPSP January 2011 The CDN surrogates can act as the super-nodes of different P2P streaming vendors with PPSP. The interactions among these network entities are the same as shown in Figure 2. The P2P streaming vendors can rent CDN resources to provide better QoS assured services for VIP users than services provides by only ordinary peers. Another case is that the CDN providers offer P2P streaming distribution with PPSP. This often occurs for an operator or CDN vendor to build a new CDN system supporting streaming applications with low cost. This service is very useful for a small streaming provider who has no much money to distribute its stream worldwide. It can also be used by an operator to launch self-operating streaming service from the beginning. 5.3. PPSP supporting cross-screen streaming in heterogeneous environment In this scenario PC, Setbox/TV and mobile terminals from both fixed network and mobile network work together sharing the content they store/cache and finishing the streaming delivery. Using PPSP, peers can identify the types of networks, peer abilities and and get to know what content other peers have (maybe with the conversion of the content availability expression in different networks) even in different network conditions as shown in Figure 3. This will play an important role on the sharing among heterogeneous peers. Zhang Expires July 16, 2011 [Page 14] Internet-Draft Problem Statement of PPSP January 2011 +-------------------------------------------------------------------+ | | | Tracker Protocol +---------+ Tracker Protocol | | +-------------> | Tracker |<------------------+ | | | +---------+ | | | | ^ | | | | | | | | | | | | | V | V | | +------+ | +------------+ | | | STB | Tracker Protocol |Mobile Phone| | | +------+ | +------------+ | | ^ | ^ | | | | | | | | | | | | | V | | | |Peer Protocol +---------+ Peer Protocol | | | +-------------> | PC |<------------------+ | | +---------+ | | | +-------------------------------------------------------------------+ Figure 3 Heterogeneous P2P Streaming Interactions with PPSP 5.4. Supporting P2P streaming in cellular mobile network In a cellular mobile environment like 3G or 4G, with the increase in bandwidth and smart mobile terminal capabilities, P2P streaming is easier to be realized than before. Note that the mobile terminals are not compulsorily to be peers. Network peers who are deployed by the ISPs or operators and mobile peers with WiFi connections are more often selected. For example, in 3GPP, there is a P2P CDS work item on the requirement of mobile operators to prefer use deployed network- side equipments (e.g., serving gateways or GGSNs, one access point from cellular mobile network to the Internet) to act as super-peers when there are no enough eligible peers to realize P2P streaming[P2P CDS]. Because they are deployed by the operators, the stability and storage size are better guaranteed than ordinary peers. In this case the PPSP tracker protocol will identify the network types and dynamics as well as the terminal types and return super- peers in the peer-list to these super-peers and normal mobile peers. If mobile terminals are not eligible to be peers, they can simply receive data from these super-peers without contributing any data to others. Zhang Expires July 16, 2011 [Page 15] Internet-Draft Problem Statement of PPSP January 2011 5.5. Cache service supporting P2P streaming As discussed in the section3, deploying cache nodes in the network edges can greatly decrease the inter-network traffic and increase user experience in streaming service. With PPSP, the cache nodes can identify the P2P streaming genre even it may include different applications, cache the frequent visited content (or part of) and report what they cache to the provider's tracker like a normal peer and serve other requesting peers in sharing data as shown in Figure 4. The cache nodes needn't update their library when new applications are introduced, which enable the cache nods spend less cost to support more applications. +-------------------------------------------------------------------+ | | | Tracker Protocol +---------+ Tracker Protocol | | +-------------> | Tracker |<--------------------+ | | | +---------+ | | | | | | | | | | | | | | | V V | | +-----------+ Peer Protocol +------------+ | | | Cache |<------------------------------->| Peer | | | +-----------+ +------------+ | +-------------------------------------------------------------------+ Figure 4 Cache Service Supporting Streaming with PPSP Zhang Expires July 16, 2011 [Page 16] Internet-Draft Problem Statement of PPSP January 2011 6. Security Considerations PPSP will not attempt to provide a solution on security and copyright issues like malicious content distribution, content pollution and DRM for a general P2P streaming system. Instead PPSP security considerations involve the security problems related to PPSP protocols. 6.1. Tracker Protocol Malicious peers may issue denial of service attack to the trackers by sending large amount of requests with tracker protocol. Distributed trackers deployment may alleviate the problem. On the other hand, malicious peers may report fake information (e.g., cheating trackers and other peers by claiming itself owning some unexisting data). So it may be optional in some cases to realize authentication to the peers before accepting the request for the tracker. But this may add up the tracker's workload on authentication. In the above discussion tracker is trustful. Things are worse when the malicious peer acts as part of the distributed trackers, who is untrustful with much possibility. The malicious acting tracker may reply the peers with fake peer-list. Peers may find they cannot find desired data with the fake peer-list. 6.2. Peer Protocol Similar to the behavior in the tracker-peer interaction, malicious peers may also create fake information on chunk availability and exchange it with other peers. Some techniques to check the data integrity (e.g., using checksum) may be useful for detecting the data. But this part is out of scope of PPSP. The protocol documents will contain a complete description on the security/privacy issues relevant to any usage of PPSP. Zhang Expires July 16, 2011 [Page 17] Internet-Draft Problem Statement of PPSP January 2011 7. Acknowledgements We would like to acknowledge the following people who provided review, feedback and suggestions to this document: M. Stiemerling;D. Bryan E. Marocco; V. Gurbani; R. Even; H. Zhang; C. Schmidt;L. Xiao; C. Williams; V. Pasual; D. Zhang; J. Lei. Zhang Expires July 16, 2011 [Page 18] Internet-Draft Problem Statement of PPSP January 2011 8. References [Cisco] Cisco Visual Networking Index: Forecast and Methodology, 2009-2014, http://www.cisco.com/en/US/solutions/collateral/ns341/ns525/ns537/ns7 05/ns827/white_paper_c11- 481360_ns827_Networking_Solutions_White_Paper.html [PPLive] www.pplive.com [VoD]Challenges, Design and Analysis of a Large-scale P2P-VoD System,Yan Huang et al, Sigcomm08. [CNN] www.cnn.com [PPStream] www.ppstream.com [UUSee] www.uusee.com [CNTV] www.cntv.com [Cirrus] labs.adobe.com/technologies/cirrus/ [Akamai] Understanding hybrid CDN-P2P: why limelight needs its own Red Swoosh, C. Huang et al,Proceedings of the 18th International Workshop on Network and Operating Systems Support for Digital Audio and Video, May 28-30, 2008, Braunschweig, Germany. [ChinaCache]www.chinacache.com [ComCast]http://www.afterdawn.com/news/article.cfm/2008/05/20/comcast _invests_in_p2p_streaming_startup [CCN] Content-Centric Networking,V. Jacobson, https://wiki.tools.isoc.org/@api/deki/files/2634//=1.vj.isoc.mar10.pd f [DONA]A Data-Oriented (and Beyond) Network Architecture, T. Koponen et al, Sigcomm 2007. [Van] A New Way to Look at Networking,Van Jacobson, http://video.google.com/videoplay?docid=-6972678839686672840 [RayV]www.rayv.com Zhang Expires July 16, 2011 [Page 19] Internet-Draft Problem Statement of PPSP January 2011 [Forcetech]http://www.forcetech.net/english/solutions [CDN+P2P]Efficient Large-scale Content Distribution with Combination of CDN and P2P Networks, H. Jiang et al, International Journal of Hybrid Information Technology, Vol.2, No.2, April, 2009. [RFC 5693], Application-Layer Traffic Optimization (ALTO) Problem Statement, E. Marocco et al, http://datatracker.ietf.org/doc/rfc5693/ [HPTP] HPTP: Relieving the Tension between ISPs and P2P, Guobin Shen et al, IPTPS 2007. [GENI] www.geni.net [FIND]www.nets-find.net [Statistics] http://labs.chinamobile.com/news/48283 [P2P CDS] 3GPP TR 22.906, Study on IMS based peer-to-peer content distribution services,http://www.3gpp.org/ftp/Specs/html- info/22906.htm Author's Addresses Yunfei Zhang China Mobile Communication Corporation zhangyunfei@chinamobile.com Ning Zong Huawei Technologies Co., Ltd. zongning@huawei.com Gonzalo Camarillo Ericsson Zhang Expires July 16, 2011 [Page 20] Internet-Draft Problem Statement of PPSP January 2011 Gonzalo.Camarillo@ericsson.com James Seng PPLive james.seng@pplive.com Richard Yang Yale University yry@cs.yale.edu Zhang Expires July 16, 2011 [Page 21]