idnits 2.17.1 draft-kamei-p2p-experiments-japan-07.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (June 30, 2012) is 4318 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- No issues found here. Summary: 0 errors (**), 0 flaws (~~), 1 warning (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 P2PRG S. Kamei 3 Internet-Draft NTT Corporation 4 Intended status: Informational T. Momose 5 Expires: January 1, 2013 Cisco Systems 6 T. Inoue 7 T. Nishitani 8 NTT Communications 9 June 30, 2012 11 ALTO-Like Activities and Experiments in P2P Network Experiment Council 12 draft-kamei-p2p-experiments-japan-07 14 Abstract 16 This document introduces experiments to clarify how ALTO-like 17 approach was effective to reduce network traffic made by a Council in 18 Japan to harmonize P2P technology with the infrastructure. And this 19 also provides some suggestions that might be useful for ALTO 20 architecture learned through our experiments. 22 Status of this Memo 24 This Internet-Draft is submitted to IETF in full conformance with the 25 provisions of BCP 78 and BCP 79. 27 Internet-Drafts are working documents of the Internet Engineering 28 Task Force (IETF), its areas, and its working groups. Note that 29 other groups may also distribute working documents as Internet- 30 Drafts. 32 Internet-Drafts are draft documents valid for a maximum of six months 33 and may be updated, replaced, or obsoleted by other documents at any 34 time. It is inappropriate to use Internet-Drafts as reference 35 material or to cite them other than as "work in progress." 37 The list of current Internet-Drafts can be accessed at 38 http://www.ietf.org/ietf/1id-abstracts.txt. 40 The list of Internet-Draft Shadow Directories can be accessed at 41 http://www.ietf.org/shadow.html. 43 This Internet-Draft will expire on January 1, 2013. 45 Copyright Notice 47 Copyright (c) 2012 IETF Trust and the persons identified as the 48 document authors. All rights reserved. 50 This document is subject to BCP 78 and the IETF Trust's Legal 51 Provisions Relating to IETF Documents 52 (http://trustee.ietf.org/license-info) in effect on the date of 53 publication of this document. Please review these documents 54 carefully, as they describe your rights and restrictions with respect 55 to this document. Code Components extracted from this document must 56 include Simplified BSD License text as described in Section 4.e of 57 the Trust Legal Provisions and are provided without warranty as 58 described in the Simplified BSD License. 60 Table of Contents 62 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 63 2. Background in Japan . . . . . . . . . . . . . . . . . . . . . 3 64 2.1. P2P traffic . . . . . . . . . . . . . . . . . . . . . . . 3 65 2.2. Impact on network infrastructure . . . . . . . . . . . . . 3 66 2.3. The object of P2P Network Experiment Council . . . . . . . 4 67 3. The details of the experiments . . . . . . . . . . . . . . . . 5 68 3.1. Dummy Node . . . . . . . . . . . . . . . . . . . . . . . . 5 69 4. Hint Server ('08) . . . . . . . . . . . . . . . . . . . . . . 7 70 5. High-Level Trial Results . . . . . . . . . . . . . . . . . . . 11 71 5.1. Peer Selection with P2P . . . . . . . . . . . . . . . . . 11 72 5.2. Peer Selection with the Hint Server . . . . . . . . . . . 12 73 6. Considerations . . . . . . . . . . . . . . . . . . . . . . . . 12 74 6.1. Next steps . . . . . . . . . . . . . . . . . . . . . . . . 13 75 6.2. Feedback to ALTO WG . . . . . . . . . . . . . . . . . . . 13 76 6.2.1. Hierarchical architecture for ALTO servers . . . . . . 13 77 6.2.2. Measurement mechanism . . . . . . . . . . . . . . . . 14 78 7. Security Considerations . . . . . . . . . . . . . . . . . . . 14 79 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 14 80 9. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 14 81 10. Informative References . . . . . . . . . . . . . . . . . . . . 14 82 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 15 84 1. Introduction 86 An overlay network, which is used by P2P and other applications, 87 offers the advantage of allowing flexible provision of services while 88 hiding the lower layer network. The downside is that inefficient 89 routes are often taken in the lower IP network, thereby increasing 90 the network load. Several proposals have been made to build an 91 overlay network that takes account of the information about the lower 92 layer network. [1] [2] Since the management of the Internet is highly 93 distributed, it is difficult to implement such proposals and thus 94 optimize a network without the cooperation of network providers. 96 Recently, the controversy between the overlay network and the network 97 providers has been rekindled. Under these circumstances, some 98 researchers have studied overlay network control technology that 99 takes account of the network topology information obtained from 100 network providers. 102 One of the activities concerning this issue has been made by the P2P 103 Network Experiment Council in Japan. This document reports on the 104 issues addressed and experiments being made by the council, focusing 105 on the experiments made from 2007 to 2008. 107 2. Background in Japan 109 2.1. P2P traffic 111 As of 2008, the world most popular P2P file sharing application, 112 Bittorrent, isn't widely deployed in Japan. Instead, other Japan 113 specific file sharing P2P applications such as Winny [3], Share [4], 114 and so on, still occupy 40% of the Internet traffic in Japan even 115 though many those P2P users were arrested for sharing illegal files 116 with these P2P apps. 118 Each P2P file sharing application has a unique protocol and none of 119 them have a large market share therefore making it hard to 120 effectively control. 122 2.2. Impact on network infrastructure 124 One of the advantage of using P2P technology for content delivery is 125 that peers exchange content directly among themselves. This reduces 126 the load on servers. Also, P2P applications can reduce upstream 127 traffic from an original content server. This is significant that 128 the charge for upstream traffic is usually delay-sensitive for 129 content delivery services, and it is not negligible. 131 It is also known that server cost could be reduced with P2P 132 technology. However, the story is quite different for network 133 providers. From the viewpoint of network providers, the traffic that 134 content servers generate has shifted to the edge network and the 135 amount of traffic has not necessarily been reduced with using P2P 136 technology for reducing server cost. Another problem for network 137 providers that an extremely inefficient routing may be selected has 138 been raised. It is because overlay network systems are configured 139 without any regard to the structure of the lower layer network or 140 network geometry. 142 In some cases, the total amount of traffic on the Internet used to be 143 limited by the capacity of servers. For those cases, P2P technology 144 can improve the scalability of servers , however it may exhaust 145 network resources. Moreover, using P2P applications increases the 146 volume of traffic per user remarkably. 148 Faced with increase in the load on network infrastructure, network 149 providers are compelled to take actions to overcome the sudden 150 increase in facilities' cost. Representative actions include placing 151 content in internet exchanges or data centers, introducing bandwidth 152 control, and raising the access fees [5]. 154 In the future, video posting sites, which has been delivered using 155 client-server applications, may adopt P2P system. The increase in 156 traffic arising from such a shift will be a great threat to the 157 network. 159 2.3. The object of P2P Network Experiment Council 161 In order to reduce Internet traffic and encourage legitimate use of 162 P2P technologies, the Japanese government led to establish a new 163 council called P2P Network Experiment Council conjunction with 164 commercial P2P application vendors and ISPs in 2006. 166 Then the council had started to develop regulations that include 167 several guidelines such like an advance notice to restrict bandwidth 168 to heavy traffic users. In accordance with the regulations, some 169 ISPs introduced solutions that reduce traffic caused by P2P file 170 sharing applications. 172 Besides this activity, the council also looked for new ways to 173 control traffic by commercial P2P applications with ISPs, carriers, 174 contents providers and P2P system vendors. In this work, the council 175 had experiments that introduced ALTO-like system and observed how the 176 traffic was reduced by redirecting to proper peers on the real 177 Internet in Japan. 179 In our experiment, the council settled hint servers, which are 180 described in section 4. Hint servers have a protocol offering 181 network distance to peers, which is disclosed to P2P application 182 vendors. 184 Using hint server, P2P application vendors can introduce ALTO 185 concepts easily to their P2P distribution systems. Because the 186 protocol provided of hint servers is independent on specific P2P 187 application vendors like Bittorrent, the council defines the protocol 188 to be able to use any P2P application vendors. It needs to gather 189 network information from ISPs to offer network distance to peers, 190 however many ISPs dislike to disclose such information to others. 191 Therefore, hint servers are designed to offer little information 192 about ISPs' network architecture to P2P application vendors. 194 To monitor traffic of peers, the council also settled dummy node, 195 which are described in section 3.1. 197 This memo describes the overview of the experiments. 199 3. The details of the experiments 201 The council has already learned that the server cost could be reduced 202 with using P2P technology for contents delivering by investigating 203 data offered by the members of the council. For example, the data 204 brought by the vendors shows as follows: 206 90% of traffic was reduced with UG Live by Utagoe Inc [6]. 208 The costs of delivering to tens of thousand subscribers was 209 reduced to 1/5 with BBbroadcast with TV Bank Corp. [7] 211 On the other hand, these reduced server costs may affect network 212 load. One of the goals of our experiments is to visualize the 213 impacts and propose an architecture to reduce network load caused by 214 these new technologies. 216 To satisfy the above goals, the framework to be proposed should be 217 well generalized as possible that doesn't rely specific P2P 218 application behaviors because multi P2P application vendors join 219 these experiments. In addition, the traffic should be captured 220 beyond multi ISPs. 222 3.1. Dummy Node 224 As mentioned before, while the effect of delivery using P2P 225 technology on reducing the traffic and the load on servers is well 226 known, traffic behavior in the inter-ISP is not known. In Japan, 227 there is a backbone traffic report cooperated with ISPs and IXes [8]. 228 However, this measurement requires to capture packets on subscribers 229 line to know end user's activity. It is not realistic to measure the 230 behavior of P2P applications at user terminals connected to the 231 Internet because that would require a large-scale arrangement for 232 measurement, such as using Deep Packet Inspection (DPI) on aggregated 233 lines. 235 To solve these problems, we put several nodes called 'dummy nodes' in 236 the ISP's networks. The dummy nodes emulate an end user's PC and P2P 237 applications are running on the nodes. 239 By introducing dummy nodes, we can observe and evaluate how much P2P 240 applications have affected networks by measuring the traffic on dummy 241 nodes. Since this method can't measure every subscriber's traffic, 242 the accuracy would be less than other methods. But this make it 243 possible to adapt to situations many different P2P applications 244 coexist on a network. We can say this is suitable for these 245 experiments. 247 A dummy node consists of Intel PC server, Linux(CentOS), VMWare and 248 Windows XP works on VMWare. With this configuration, all packets can 249 be captured without any impacts to the network, nodes and application 250 behaviors. And it enable us to use different P2P applications for 251 windows and evaluate them generally. 253 To see behaviors of the node, incoming and outgoing packets are 254 captured on Linux because every packets are transmitted through it. 255 In these experiments, we captured source/destination address, port 256 number, amount of traffic and start/end time to see flow information. 258 60 Dummy nodes are put on access networks that are closest subscriber 259 as possible in different 40 networks. 261 +----------------------+ 262 |+--------------------+| 263 ||+------------------+|| 264 ||| P2P Application ||| 265 ||| WindowsXP ||| 266 ||| +--+ ||| 267 ||+--------|N |------+|| 268 || VMware |e | || 269 |+---------|t |-------+| 270 | Linux |IF| capture| 271 +----------| |--------+ 272 +--+ 274 Dummy nodes 276 Figure 1 278 4. Hint Server ('08) 280 In Japan, bottleneck in IP networks have been shifting from access 281 networks to backbone networks and equipments, such as bandwidth 282 between ISPs and capacity in IXs, since FTTH has rapidly spread all 283 over Japan. Under these circumstances, the Council proposed a less 284 restrictive and more flexible cooperation between ISPs than existent 285 P4P experiments [9]. The proposed method consists of the following 286 elements: (1) P2P clients, (2) P2P control servers, and (3) a hint 287 server: a peer selection hint server. (1) and (2) are existing 288 systems but whether (2) exists depends on each application. (3) is a 289 server that provides a hint as to the selection of a peer, and plays 290 a role equivalent to that of ALTO Server. Note that this proposal 291 was based on results of experiments using dummy nodes. The results 292 showed that it was possible to reduce unnecessary traffic that flows 293 across the boundaries of geographical districts or ISPs through 294 providing information about the physical network to P2P applications. 296 When a peer joins the network, it registers its location information 297 (IP address) and supplementary information (line speed, etc.) with&# 298 12288;the hint server. The hint server calculate network distance 299 between peers (P2P client) based on network topology information 300 obtained from the ISP and generates a priority table for peer 301 selection. The hint server returns the table to the peer. 303 If all information can be made publicly, the above procedure can 304 produce a result which is close to overall optimization. However, 305 some information held by ISPs can often be confidential. Besides, in 306 some cases, the volume of calculation required to process all 307 information can be excessive. To avoid these problems, it is planned 308 to conduct experiments with a limited set of functions, analyze 309 experiments results, and gradually expand the scope of optimization. 311 A control mechanism that makes use of all possible information is 312 difficult not only technically but also difficulties to achieve 313 coordination among providers. In consideration of these 314 difficulties, the council has been limiting the implementation and 315 experiments to the following scope since 2006. 317 Figure 2 shows an outline of the hint server. 319 +---------+ GetLocation +-------------GeoIP DB Server---------+ 320 | | +-----------+ | +----------+ +-----------+ | 321 | |--|IP Address |-->| | GeoIP DB | |Quagga etc | | 322 | | +-----------+ | +----------+ +-----------+ | 323 | | | +-------------+ +----------------+ | 324 | | +-----------+ | | District | | Routing | | 325 | |--|AS Code: |---| | information | |information(DGP)| | 326 | | |Regional | | | | | | | 327 |P2P Peers| |Information| | | Range of | |AS Code(origin) | | 328 | or | +-----------+ | | IP address | | | | 329 | Contro| | | +-------------+ +----------------+ | 330 | Server | +-------------------------------------+ 331 | | | ^ 332 | | PeerSelection v | 333 | | +-----------+ +--------------------------------------+ 334 | |--|IP Address |-->| +--Priority Node Selection System--+ | 335 | | | List | | | | | 336 | | +-----------+ | | Peer candidate ranking | | 337 | | +-----------+ | | | | 338 | |--| Ranking |-->| +----------------------------------+ | 339 | | +-----------+ +--------------------------------------+ 340 +---------+ 342 Peer selection hint server 344 Figure 2 346 The network information used by the hint server is not information 347 solicited from individual ISPs but the AS number and district 348 information, which are more or less already public. Routing tables 349 are not generated. Instead, peers within the same ISP or the same 350 district are selected with higher priority in order to confine 351 traffic to within the same ISP or the same district. 353 When the hint server receives an IP address, it returns its attribute 354 information, to achieve the above. A peer can select a peer based on 355 the returned information. This operation is called GetLocation. 356 However, in preparation for the time when it becomes necessary to 357 hide topology information, an interface is provided through which a 358 priority order is returned in response to an input of a list of 359 candidate peers. This operation is called PeerSelection. 361 Although the target node is selected based on the criterion that it 362 is within the same ISP or the same district, this type of selection 363 is not very effective if the number of participating peers is small. 364 Table 1 shows ratio of peers within the same AS or the same 365 prefecture calculated from the distribution of ASs and prefectures in 366 the IP address space from one-day data on a Winny network. 368 +--------------------+--------+ 369 | Conditions | ratio | 370 +--------------------+--------+ 371 | AS matches | 6.70% | 372 | Prefecture matches | 12.76% | 373 | Both match | 2.09% | 374 | Neither match | 78.45% | 375 +--------------------+--------+ 377 Table 1: AS and prefecture distributions 379 Since, in addition to the above, the presence/absence of content 380 affects the result, the control of selecting a peer within the same 381 district may be inadequate. Therefore, it is necessary to introduce 382 the weight of a continuous quantity that reflects the physical 383 distance or the AS path length as an indicator of the proximity of 384 the areas involved. 386 In consideration of the above, the following two measures are used 387 for the evaluation of proximity between peers in a hint server. 389 o AS path length (distance between ISPs) 391 AS path length calculated from BGP full routes. Since a full 392 routing table retrieved at an ISP can show only a best path, it 393 may not get an accurate length if the AS hop of both ISPs is too 394 large. To avoid this, we use multiple BGP information received 395 from different ISPs and combine them. Based on this concept, we 396 used BGP routing information's offered by three ISPs operated by 397 big telecommunication couriers and made a topology tree. Then it 398 enables to calculate the shortest path between given two ASes. 400 o Geographical distance 401 Distances between peers are measured using physical distance of 402 prefectural capitals that target peers belong to. The distance 403 between prefectural capitals is used to calculate physical 404 distance. Distances between prefectural capitals are sorted into 405 ascending order, and then into bands, with weights 1 to 15 406 assigned to them so that there are a more or less equal number of 407 "capital pairs" in each band. If either of their location is 408 indefinite, distance is equal to 15 and, if they are in the same 409 prefecture, distance is equal to 0. 411 Evaluation of distances between peers showed that the distribution 412 of distances was almost uniform when distances between peers are 413 normalized. This result suggests that using normalized distances 414 expands the area where the control by a Hint Server is effective. 415 The geographical distance is only used when the AS path length is 416 same. 418 An example of the request and the response 420 o Request 422 POST /PeerSelection HTTP/1.1 423 Host: ServerName 424 User-Agent: ClientName 425 Content-Type: text/plain; charset=utf-8 427 v=Version number 428 [application=Application identifier] 429 ip=IP address of physical interface 430 port=Port number of physical interface 431 [nat={no|upnp|unknown}] 432 [nat_ip=Global IP address using UPnP] 433 [nat_port= Global port number using UPnP] 434 [trans_id=transcation ID] 435 [pt=Flag of port type] 436 [ub=upload bandwidth] 437 [db=download bandwidth] 439 o Response 441 HTTP/1.1 200 OK 442 Date: Timestamp 443 Content-Type: text/plain; charset=utf-8 444 Cache-control: max-age=max age 445 Connection: close 447 v=Version number 448 ttl=ttl 449 server=hint server name 450 ... 451 trans_id=transaction ID 452 pt=Flag of port type 453 client_ip=Peer IP address observed from server 454 client_port=Peer port number observed from server 455 numpeers=number of respond peer 456 n=[src address] dst address / cost / option 458 5. High-Level Trial Results 460 5.1. Peer Selection with P2P 462 Table 2 shows the result of the analysis of communication in a node 463 of an ISP installed in Tokyo, as an example of measurement results. 465 In these two experiments we evaluate different P2P applications, in 466 1st experiment, the P2P topology is generated by tree algorithm, and 467 in 2nd experiment, it is generated by mesh algorithm. Both of them 468 result in similar performance. 470 +-----------------------------------------+------------+------------+ 471 | Conditions | Experiment | Experiment | 472 | | 1 | 2 | 473 +-----------------------------------------+------------+------------+ 474 | *Peers selected within the same ISP | 22% | 29% | 475 | *Peers selected within the same | 19% | 23% | 476 | district | | | 477 | *Peers selected within the same | 5% | 7% | 478 | district and the same ISP | | | 479 +-----------------------------------------+------------+------------+ 481 Table 2: Percentage of communication within the same ISP 483 The table shows that the probability of communication with peers in 484 the same ISP is proportional to the number of population and the 485 share of the ISP in each district. The data show that peers were 486 selected at random. Note that the vendor of a P2P application used 487 in these experiments explained that the mechanism of selection a peer 488 using network information can be implemented. However, peer 489 selection is normally based on past information because users often 490 cannot actually perceive the effect of using network information. 492 5.2. Peer Selection with the Hint Server 494 The main objective of these experiments was to verify the operations 495 of the hint server and P2P applications. The distances between a 496 dummy node and a peer were obtained from data on the dummy nodes. An 497 examination of the distances between a dummy node and a peer revealed 498 that mean value of distance after the hint server was introduced was 499 reduced by 10% and that 95% value of that was reduced by 5%. The 500 results show introducing hint server can reduce network loads by P2P 501 applications. 503 6. Considerations 505 We clarified followings throughout our experiments. 507 1. Dispersed dummy nodes can figure out the behavior of peers and 508 traffic between inter-ISP networks, which peers are selected by 509 each peer. Therefore it proves that the importance of peer 510 selection control mechanism proposed in ALTO. 512 2. Using our peer selection control mechanism, called hint server, 513 could achieve significant differences. Our hint server can lead 514 each peer to select nearer peer. 516 In the experimental result of peer selection control, it is smaller 517 in intra-ISP traffic than other experiments [10] We think that it is 518 because there are smaller peers in each area of traffic control. 519 When there are many peers in one ISP, it is easy to select peers in 520 the same ISP. However, when there are small peers in one ISP, it is 521 difficult to select peers in the same ISP. In the situation of our 522 experiments, there are many ISPs of peers belonging, and there are 523 relatively smaller peers exist in same ISP. 525 Moreover, we didn't force P2P vendors to limit their implementation 526 policy, therefore we can observe differences how each implementations 527 weigh the information from the hint servers. Especially, in tree 528 overlay topology P2P applications, such mechanism is very effective, 529 on the other hand, in mesh overlay system, less effective. 531 6.1. Next steps 533 The experiments are on going as of 2011. Current experiments in 534 2011, we've changed the communication protocol to hint servers to 535 ALTO based because it is nearly standardized. In our implementation, 536 PIDs and the value of cost are mapped to ISP subnets, and ISP 537 distance respectively. We also implement services for compatibility 538 required by ALTO such as Service Capability and Map Services. But 539 the Endpoint Cost Service is mainly used because of backward 540 compatibility of our experiments. 542 We also study hierarchical hint server structure, in order to control 543 in coarse inter-ISPs and in detail intra-ISP. It is also effective 544 for limiting the area of information disclose. 546 6.2. Feedback to ALTO WG 548 This section describes what the authors learned with these 549 experiments would be useful for the ALTO WG. 551 6.2.1. Hierarchical architecture for ALTO servers 553 In our experiments, we present the possibility of traffic control 554 among multi-ISPs and multi-P2P applications using ALTO mechanism. On 555 the other hand, we found several problems in ISP operations to adapt 556 the mechanism. One is the granularity of network information. Among 557 inter-ISP area, it is relatively easy to treat information for public 558 purpose using BGP full route. On the other hand, among intra-ISP 559 area, it may be difficult to disclose private information of each 560 ISP. [11] propose some modification for ALTO protocol in order to 561 hide ISP information. We propose hierarchical structures. From the 562 viewpoint of cooperation between ISPs, fine-grained information is 563 not necessarily required and moreover it is difficult to exchange the 564 fine-grained information between ISPs. Considering this situation, 565 the authors use only coarse-grained information to control backbone 566 traffic in the experiments this year, though demand of controlling 567 traffic within an ISP using fine-grained information will arise in 568 the near future. Therefore it led us that introducing hierarchical 569 structure into ALTO is necessary to cope with both situations. 570 Actually, the authors plan to adapt a hierarchical control mechanism 571 in the next steps, which include the following two steps. 573 o In the first step, coarse-grained information about whole the 574 network is used to select ISPs. 576 o Next, fine-grained information within the ISP is used to select a 577 peer. 579 6.2.2. Measurement mechanism 581 In the experiments, there were two difficulties as follows: 583 o Evaluating effect of introducing a hint server was difficult, 584 since P2P applications had their own measurement mechanisms. 586 o How to treat priority orders of peers suggested by a hint server 587 could not be predetermined for P2P applications. 589 From these experiences, the authors consider that clarifying 590 requirements about measurement mechanisms for P2P applications are 591 necessary also in ALTO. 593 7. Security Considerations 595 This document does not propose any kind of protocol, practice or 596 standard. 598 8. IANA Considerations 600 No need to describe any request regarding number assignment. 602 9. Acknowledgments 604 Thanks to strong support by MIC (Ministry of Internal Affairs and 605 Communications of Japanese government), the council was established. 606 These experiments were performed under cooperation among P2P Network 607 Experiment Council members, and DREAMBOAT co.,ltd., Bitmedia Inc., 608 Utagoe. Inc. and Toyama IX have especially supported analyses of the 609 experiments. The authors appreciate Tohru Asami, Hiroshi Esaki and 610 Tatsuya Yamshita for their constructive comments. 612 10. Informative References 614 [1] "On the Quality of Triangle Inequality Violation Aware Routing 615 Overlay Architecture", INFOCOM 2009: 2761-2765. 617 [2] "QRON: QoS-aware routing in overlay networks", IEEE JOURNAL ON 618 SELECTED AREAS IN COMMUNICATIONS, VOL. 22, NO. 1, JANUARY 2004. 620 [3] "Winny on Wikipedia", . 622 [4] "Share on Wikipedia", 623 . 625 [5] Taniwaki, "Broadband Competition Policy in Japan", 2008, 626 . 628 [6] Utagoe Inc., "UGLive technology introduction", 629 http://www.utagoe.com/en/technology/grid/live/index.html, 630 March, 2011. 632 [7] TVBank, "Live Delivery using `BB Broadcast'Achieving 96% Saving 633 in Traffic!", http:.wwww.tv-bank.com/jp/20081031.html, 2008 (in 634 Japanese). 636 [8] Cho, Fukuda, Esaki, and Kato, "The Impact and Implications of 637 the Growth in Residential User-to-User Traffic", SIGCOMM2006, 638 pp207-218, Pisa, Italy, September 2006. 640 [9] Open P4P, "P4P Field Tests: Yale-Pando-Verizon", 641 http://www.openp4p.net/front/, 2009. 643 [10] "RFC5632: Comcast's ISP Experiences in a Proactive Network 644 Provider Participation for P2P (P4P) Technical Trial", 645 September 2009. 647 [11] "ALTO H12,draft-kiesel-alto-h12-02 (work in progress)", March 648 2010. 650 Authors' Addresses 652 Satoshi Kamei 653 NTT Service Integration Laboratories 654 3-9-11, Midori-cho 655 Musashino-shi, Tokyo 180-8585 656 JP 658 Phone: +81-422-59-6942 659 Email: kamei.satoshi@lab.ntt.co.jp 660 Tsuyoshi Momose 661 Cisco Systems G.K. 662 9-7-1 Akasaka 663 Minato-ku, Tokyo 107-6227 664 JP 666 Phone: +81-3-6738-5154 667 Email: tmomose@cisco.com 669 Takeshi Inoue 670 NTT Communications 671 3-4-1, Shibaura 672 Minato-ku, Tokyo 108-8118 673 JP 675 Phone: +81-3-6733-7177 676 Email: inoue@jp.ntt.net 678 Tomohiro Nishitani 679 NTT Communications 680 1-1-6, Uchisaiwaicho 681 Chiyodaku, Tokyo 100-8019 682 JP 684 Phone: +81-50-3812-4742 685 Email: tomohiro.nishitani@ntt.com