P2PRG S. Kamei
Internet-Draft NTT Corporation
Intended status: Informational T. Momose
Expires: May 24, 2011 Cisco Systems
T. Inoue
T. Nishitani
NTT Communications
November 20, 2010
ALTO-Like Activities and Experiments in P2P Network Experiment Council
draft-kamei-p2p-experiments-japan-04
Abstract
This document provides some suggestions about ALTO architecture
through experiments made by P2P Network Experiment Council in Japan.
This document also introduces experiments made by the Council in
Japan to harmonize P2P technology with the infrastructure.
Specifically, this document describes Hint Server technology, which
is similar to ALTO technology.
Status of this Memo
This Internet-Draft is submitted to IETF in full conformance with the
provisions of BCP 78 and BCP 79.
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.
The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html.
This Internet-Draft will expire on May 24, 2011.
Copyright Notice
Copyright (c) 2010 IETF Trust and the persons identified as the
Kamei, et al. Expires May 24, 2011 [Page 1]
Internet-Draft P2P Experiments Japan November 2010
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 BSD License.
Table of Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3
2. Background in Japan . . . . . . . . . . . . . . . . . . . . . 3
2.1. P2P traffic . . . . . . . . . . . . . . . . . . . . . . . 3
2.2. Impact on network infrastructure . . . . . . . . . . . . . 4
3. Activity in P2P Network Experiment Council . . . . . . . . . . 5
3.1. Dummy Node . . . . . . . . . . . . . . . . . . . . . . . . 5
3.2. Hint Server ('08) . . . . . . . . . . . . . . . . . . . . 5
3.3. Difference between P4P and Hint Server technology . . . . 9
3.4. Difference between ALTO and Hint Server technology . . . . 11
4. High-Level Trial Results . . . . . . . . . . . . . . . . . . . 11
4.1. Peer Selection with P2P . . . . . . . . . . . . . . . . . 11
4.2. Peer Selection with the Hint Server . . . . . . . . . . . 12
5. Next steps . . . . . . . . . . . . . . . . . . . . . . . . . . 12
6. Feedback to ALTO WG . . . . . . . . . . . . . . . . . . . . . 13
6.1. Harmonizing a Hint Server with ALTO . . . . . . . . . . . 13
6.2. Measurement mechanism . . . . . . . . . . . . . . . . . . 13
7. Security Considerations . . . . . . . . . . . . . . . . . . . 14
8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 14
9. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 14
10. Informative References . . . . . . . . . . . . . . . . . . . . 14
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 15
Kamei, et al. Expires May 24, 2011 [Page 2]
Internet-Draft P2P Experiments Japan November 2010
1. Introduction
An overlay network, which is used by P2P and other applications,
offers the advantage of allowing flexible provision of services while
hiding the lower layer network. The downside is that inefficient
routes are often taken in the lower IP network, thereby increasing
the network load. Several proposals have been made to build an
overlay network that takes account of the information about the lower
layer network. Since the management of the Internet is highly
distributed, it is difficult to implement such proposals and thus
optimize a network without the cooperation of network providers.
Recently, the controversy between the overlay network and the network
providers have been rekindled. Under these circumstances, some
researchers have studied overlay network control technology that
takes account of the network topology information obtained from
network providers.
One of activities concerning this issue has been made by the P2P
Network Experiment Council in Japan. This document reports on the
issues addressed and experiments being made by the P2P Network
Experiment Council in Japan, focusing on the experiments made from
2007 to 2008.
2. Background in Japan
2.1. P2P traffic
In Japan, the major of P2P applications used today is Winny [1]. P2P
applications are the sources of a considerable volume of traffic.
Recent study [2] showed more than 60% of Internet traffic in Japan is
generated by P2P applications.
Although traffic from P2P applications increased much more rapidly
than traffic from client-server-type web applications, it has leveled
off lately as a result of legal restrictions advocated by copyright
management organizations and traffic control implemented by ISPs.
According to [3], video delivery sites using Flash has again
increased volume of web traffic per user, making P2P traffic
relatively less conspicuous than before.
Consequently, some believe that P2P traffic is no longer a threat to
the infrastructure. P2P applications, however, rapidly became widely
used to get around the limit of the servers' capacity, which was
caused by the increase in demand for delivery of music files. As of
2007, it was likely that the traffic of client-server video delivery
would shift to P2P delivery.
Kamei, et al. Expires May 24, 2011 [Page 3]
Internet-Draft P2P Experiments Japan November 2010
In fact, some P2P content delivery systems solve copyright issues,
for example, Sharecast, Ocean-Grid, TVBand, and so on. The
transmission of President Obama's Inaugural Address, which is the
largest-scale transmission of content in recent history, was mostly
of the client-server type. However, the delivery by CNN used a P2P
plug-in made by Octoshape.
2.2. Impact on network infrastructure
One of advantage of using P2P technology for content delivery is that
peers exchange content directly among themselves. This reduces the
load on servers. Also, P2P applications can reduce upstream traffic
from an original content server. This is significant that the charge
for upstream traffic is usually traffic-sensitive for content
delivery services, and it is not negligible.
Actually, the volume of traffic sent by the content server in
TVBank's P2P content delivery was reduced by a maximum of 96%
compared with the volume of traffic received by users [4]. This
indicates the great cost-saving of P2P technology from the
perspectives of the load on server hardware and the traffic relaying
cost of data centers. However, the story is quite different for
network providers. From viewpoint of network providers, the traffic
that content servers generate has shifted to the edge network and the
amount of traffic has not necessarily been reduced. Another problem
for network providers that an extremely inefficient routing may be
selected has been raised. It is because overlay network systems are
configured without any regard to the structure of the lower layer
network or network geometry.
In some cases, traffic on the Internet used to be limited by the
capacity of servers. For those cases, the improvement in the
scalability of servers has made it likely that network resources will
be used up before server resources are. Using P2P applications
increases the volume of traffic per user remarkably.
Faced with increase in the load on network infrastructure, network
providers are compelled to take actions to overcome the sudden
increase in facilities' cost. Representative actions include placing
content in IXs or data centers, introducing bandwidth control, and
raising the access fees [2].
In the future, video posting sites, which has been delivered using
client-server applications, may adopt P2P system. The increase in
traffic arising from such a shift will be a great threat to the
network.
Kamei, et al. Expires May 24, 2011 [Page 4]
Internet-Draft P2P Experiments Japan November 2010
3. Activity in P2P Network Experiment Council
3.1. Dummy Node
While the effect of delivery using P2P technology on reducing the
traffic and the load on servers is well known, traffic behavior in
the Internet is not known. However, it is not realistic to measure
the behavior of P2P applications at user terminals connected to the
Internet because that would require a large-scale arrangement for
measurement, such as using Deep Packet Inspection (DPI) on aggregated
lines. To solve this problem, dummy nodes which act like consumers
node have been introduced. Dummy nodes have been settled in the
Internet and P2P applications have been installed on these nodes.
Dummy nodes enable us to measure and analyze communication among
peers.
Specifically, Linux servers were installed at 40 sites of some ISPs,
and a virtual Windows environment was installed on the servers. P2P
applications which were target to measure were running on that
environment, and packets were captured by a Linux program to obtain
information on communication destinations and communication
frequencies. The nodes are placed on data centers, however, it
wasn't matter because the bandwidth limited the subscribers line
isn't mesured in this experiment.
3.2. Hint Server ('08)
In Japan, bottleneck in IP networks will shift from access networks
to backbone networks and equipments, such as bandwidth between ISPs
and capacity in IXs, since FTTH has rapidly spread all over Japan.
Under this situation. the Council proposed a less restrictive and
more flexible cooperation between ISPs than ALTO. The proposed
method consists of the following elements: (1) P2P clients, (2) P2P
control servers, and (3) a peer selection hint server, and a Hint
Server. (1) and (2) are existing systems but whether (2) exists
depends on each application. (3) is a server that provides a hint as
to the selection of a peer, and plays a role equivalent to that of
iTracker in P4P's study. Note that this proposal was based on
results of experiments using dummy nodes. The results showed that it
was possible to reduce unnecessary traffic that flows across the
boundaries of districts or ISPs through providing information about
the physical network to P2P applications.
When a peer joins the network, it registers its location information
(IP address) and supplementary information (line speed, etc.) with
the Hint Server. The Hint Server makes a mapping of the new peer
(P2P client) based on network topology information obtained from the
ISP, generates a routing table in which peers are listed in the order
Kamei, et al. Expires May 24, 2011 [Page 5]
Internet-Draft P2P Experiments Japan November 2010
of priority for selection, and returns the table to the peer.
If all information can be made public, the above procedure can
produce a result which is close to overall optimization. However,
some information held by ISPs can often be confidential. Besides, in
some cases, the volume of calculation required to process all
information can be excessive. To avoid these problems, it is planned
to conduct experiments with a limited set of functions, analyze
experiment results, and gradually expand the scope of optimization.
A control mechanism that makes use of all possible information is
difficult not only technically but also because it is difficult to
achieve coordination among providers. In consideration of these
difficulties, the P2P Network Experiment Council has been limiting
the implementation and experiments to the following scope since 2006.
Figure 1 shows an outline of the hint server.
+---------+ GetLocation +-------------GeoIP DB Server---------+
| | +-----------+ | +----------+ +-----------+ |
| |--|IP Address |-->| | GeoIP DB | |Quagga etc | |
| | +-----------+ | +----------+ +-----------+ |
| | | +-------------+ +----------------+ |
| | +-----------+ | | District | | Routing | |
| |--|AS Code: |---| | information | |information(DGP)| |
| | |Regional | | | | | | |
|P2P Peers| |Information| | | Range of | |AS Code(origin) | |
| or | +-----------+ | | IP address | | | |
| Contro| | | +-------------+ +----------------+ |
| Server | +-------------------------------------+
| | | ^
| | PeerSelection v |
| | +-----------+ +--------------------------------------+
| |--|IP Address |-->| +--Prioryty Node Selection System--+ |
| | | List | | | | |
| | +-----------+ | | Peer candidate ranking | |
| | +-----------+ | | | |
| |--| Ranking |-->| +----------------------------------+ |
| | +-----------+ +--------------------------------------+
+---------+
Figure 1
The network information used by the Hint Server is not information
solicited from individual ISPs but the AS number and district
information, which are more or less already public. Routing tables
are not generated. Instead, peers within the same ISP or the same
Kamei, et al. Expires May 24, 2011 [Page 6]
Internet-Draft P2P Experiments Japan November 2010
district are selected with higher priority in order to confine
traffic to within the same ISP or the same district.
When the Hint Server receives an IP address, it returns its attribute
information, to achieve the above. A peer can select a peer based on
the returned information. This operation is called GetLocation.
However, in preparation for the time when it becomes necessary to
hide topology information, an interface is provided through which a
priority order is returned in response to an input of a list of
candidate peers. This operation is called PeerSelection.
Although the priority node is selected based on the criterion that it
is within the same ISP or the same district, this type of selection
is not very effective if the number of participating peers is small.
Table 1 shows ratio of peers within the same AS or the same
prefecture calculated from the distribution of ASs and prefectures in
the IP address space from one-day data on a Winny network.
+--------------------+--------+
| Conditions | ratio |
+--------------------+--------+
| AS matches | 6.70% |
| Prefecture matches | 12.76% |
| Both match | 2.09% |
| Neither match | 78.45% |
+--------------------+--------+
Table 1: AS and prefecture distributions
Since, in addition to the above, the presence/absence of content
affects the result, the control of selecting a peer within the same
district may be inadequate. Therefore, it is necessary to introduce
the weight of a continuous quantity that reflects the physical
distance or the AS path length as an indicator of the proximity of
the areas involved.
In consideration of the above, the following two measures are used
for the evaluation of proximity between peers in a Hint Server.
o AS path length (distance between ISPs)
Distances between peers are weighted using the degree of paths'
matching from an origin AS to ASs that target peers belong to.
The degree of paths' matching means ratio of common paths from an
origin AS (for example, 4/6 between A-B-C, and A-B-D, 6/8 between
A-B-C-D and A-B-C-E). In this year, the OCN is used as an origin
AS. Distance is calculated as int((1.0- degree of matching of AS
paths)*15). Distance is 15 if either of AS path is indefinite,
Kamei, et al. Expires May 24, 2011 [Page 7]
Internet-Draft P2P Experiments Japan November 2010
and is 0 if there is a perfect match.
o Physical distance
Distances between peers are measured using physical distance of
prefectural capitals that target peers belong to. The distance
between prefectural capitals is used to calculate physical
distance. Distances between prefectural capitals are sorted into
ascending order, and then into bands, with weights 1 to 15
assigned to them so that there are a more or less equal number of
"capital pairs" in each band. If either of their location is
indefinite, distance is equal to 15 and, if they are in the same
prefecture, distance is equal to 0.
Evaluation of distances between peers showed that the distribution
of distances was almost uniform when distances between peers are
normalized. This result suggests that using normalized distances
expands the area where the control by a Hint Server is effective.
An example of the request and the response
o Request
POST /PeerSelection HTTP/1.1
Host: ServerName
User-Agent: ClientName
Content-Type: text/plain; charset=utf-8
v=Version number
[application=Application identifier]
ip=IP address of physical interface
port=Port number of physical interface
[nat={no|upnp|unknown}]
[nat_ip=Global IP address using UPnP]
[nat_port= Global port number using UPnP]
[trans_id=transcation ID]
[pt=Flag of port type]
[ub=upload bandwidth]
[db=download bandwidth]
Kamei, et al. Expires May 24, 2011 [Page 8]
Internet-Draft P2P Experiments Japan November 2010
o Response
HTTP/1.1 200 OK
Date: Timestamp
Content-Type: text/plain; charset=utf-8
Cache-control: max-age=max age
Connection: close
v=Version number
ttl=ttl
server=hint server name
...
trans_id=transaction ID
pt=Flag of port type
client_ip=Peer IP address observed from server
client_port=Peer port number observed from server
numpeers=number of respond peer
n=[src address] dst address / cost / option
3.3. Difference between P4P and Hint Server technology
To explain difference between P4P and Hint Server technology, the
architecture proposed by P4P is described. P4P aims to control
traffic in such a way that traffic is confined within the same
district or AS. As shown in Figure 2, iTracker provides an interface
for P2P content delivery using appTracker and peers in BitTorrent.
This arrangement provides a framework for efficient control based on
network information.
In this framework, it is proposed that ISPs and applications share
the following types of information through iTracker:
o Info: information about peers within an ISP
- ASID AS number
- Group number of PID node (peer)
- LOC: virtual and geographical coordinates
o Policy: information about policy on usage specified by an ISP
- Ratio between outgoing traffic and incoming traffic that flows
between domains
Kamei, et al. Expires May 24, 2011 [Page 9]
Internet-Draft P2P Experiments Japan November 2010
- Desirable daily traffic variation pattern on a link
- Specifications about relations between peer groups (PID)
o Capability: information about the capability of an ISP
- Information about usable service classes
- Information about the cache server
Note that [5] reports on the results of a field test in which it was
attempted to reduce overall traffic by using the above concept to
confine traffic exchange destinations to within the same ISP or the
same city. It reports that, in an evaluation with a Verizon network,
traffic to locations outside an ISP was reduced by 30 to 50% and that
the ratio of inter-city traffic to Verizon's total traffic was more
or less halved.
ISP
+------------------------+ Internet
| +----------------+ | +------------+
| | iTracker | | | appTracker |
| | *Info |--------> +------------+
| | *Policy | | ^
| | *capability | | |
| +----------------+ | |
| | |
| +----------------+ | |
| | Peer |----------------+
| +----------------+ |
+------------------------+
Figure 2
Comparing P4P with Hint Server technology, the following three
differences are observed:
o Target of optimization:
P4P technology focuses on optimization within an ISP, while Hint
Server technology focuses on optimization in backbone traffic.
o Target applications:
P4P technology focuses on supporting BitTorrent, while Hint Server
technology does not specify any P2P applications.
Kamei, et al. Expires May 24, 2011 [Page 10]
Internet-Draft P2P Experiments Japan November 2010
o Strength of cooperation between P2P providers and ISPs:
P4P technology requires close cooperation between ISPs and P2P
providers, while Hint Server technology does not require.
3.4. Difference between ALTO and Hint Server technology
ALTO technology is more general approach than P4P technology. And
Hint Server technology has more similar focus of this technology.
Hint Server offers similar information of ALTO service and can easily
supports following ALTO Services.
o Map Service
The cost type is computed by physical distance and AS path length,
and the mode is numerical.
PID information, it is same as AS and physical location, does not
offered to client.
o Map Filtering Service
o Endpoint Cost Service
Hint Server only offers map information associated with requested
IP addresses.
ALTO framework has more generality but the following two points are
not sufficiently improved or some operational solution should be
offered.
4. High-Level Trial Results
4.1. Peer Selection with P2P
Table 2 shows the result of the analysis of communication in a node
of an ISP installed in Tokyo, as an example of measurement results.
+-----------------------------------------+------------+------------+
| Conditions | Experiment | Experiment |
| | 1 | 2 |
+-----------------------------------------+------------+------------+
| *Peers selected within the same ISP | 22% | 29% |
| *Peers selected within the same | 19% | 23% |
| district | | |
Kamei, et al. Expires May 24, 2011 [Page 11]
Internet-Draft P2P Experiments Japan November 2010
| *Peers selected within the same | 5% | 7% |
| district and the same ISP | | |
+-----------------------------------------+------------+------------+
Table 2: Percentage of communication within the same ISP
The table shows that the probability of communication with peers in
the same ISP is proportional to the number of population and the
share of the ISP in each district. The data show that peers were
selected at random. Note that the vendor of a P2P application used
in this experiment explained that the mechanism of selection a peer
using network information can be implemented. However, peer
selection is normally based on past information because users often
cannot actually perceive the effect of using network information.
4.2. Peer Selection with the Hint Server
Since the main objective of this experiment was to verify the
operations of the Hint Server and P2P applications, the degree to
which traffic in the network was actually reduced was not evaluated.
However, the distances between a dummy node and a peer were obtained
from data on the dummy nodes. An examination of the distances
between a dummy node and a peer revealed that mean value of distance
after the Hint Server was introduced was reduced by 10% and that 95%
value of that was reduced by 5%.
5. Next steps
This document has reported on activities aimed at achieving
cooperative control between the P2P/overlay network and the network
infrastructure. Specifically, it has described issues to be
addressed and the activities of the P2P Network Experiment Council in
Japan, which was established to address these issues. It has also
introduced the Council's activities, from 2007 to 2008, focusing on
the use of a Hint Server, which is a feature of the traffic
engineering mechanism proposed by the Council.
The P2P Network Experiment Council has been renamed the Advanced
Network Use Promotion Council. The new Council aims to create new
network services suitable for the broadband environment and to
promote the widespread use of such services in rural areas. It has
expanded its scope of work to include all cache technologies,
including P2P technology. It will promote more advanced use of the
network by encouraging an exchange of views among a broad spectrum of
parties on how to use the network effectively, and by supporting a
variety of feasibility tests.
Kamei, et al. Expires May 24, 2011 [Page 12]
Internet-Draft P2P Experiments Japan November 2010
The Council aims to continue the analysis of the experiment results
obtained, and further study by involving a wider spectrum of P2P
providers, network providers and delivery service providers.
6. Feedback to ALTO WG
This section describes what the authors learned with this experiment
would be useful for the ALTO WG.
6.1. Harmonizing a Hint Server with ALTO
As described before, a Hint Server control mechanism focuses on
control between ISPs, while ALTO does control within an ISP.
Generally speaking, control mechanism that a peer chooses a replica
from its neighbors shows higher performance when probability of a
peer having a content is higher. This means ISP cooperation
mechanism that enlarges area in choosing peers will have much impact
on P2P performance. The authors consider combination of these two
mechanisms produce better P2P performance. The authors propose
hierarchical structure to harmonize a Hint Server with ALTO. From
viewpoint of cooperation between ISPs, fine information is not
necessarily required and it is difficult to exchange fine information
between ISPs. Considering this situation, the authors use only
coarse information to control backbone traffic in the experiments
this year, though demand of controlling traffic within an ISP using
fine information will arise in the near future. The authors consider
that introducing hierarchical structure into ALTO is necessary to
cope with both situations. Actually, the authors plan to try a
hierarchical control mechanism in the next steps, which include the
following two steps.
o In the first step, coarse information about whole the network is
used to select ISPs.
o Next, fine information within the ISP is used to select a peer.
6.2. Measurement mechanism
In experiments, there were two difficulties as follows:
o Evaluating effect of introducing a Hint Server was difficult,
since P2P applications had their own measurement mechanisms.
o How to treat priority orders of peers suggested by a Hint Server
could not be predetermined for P2P applications.
From these experiences, the authors consider that clarifying
Kamei, et al. Expires May 24, 2011 [Page 13]
Internet-Draft P2P Experiments Japan November 2010
requirements about measurement mechanisms for P2P applications are
necessary also in Alto.
7. Security Considerations
There are no security considerations in this document.
8. IANA Considerations
No need to describe any request regarding number assignment.
9. Acknowledgments
These experiments were performed under cooperation among P2P Network
Experiment Council members, and DREAMBOAT co.,ltd., Bitmedia Inc.,
Utagoe. Inc. and Toyama IX have especially supported analyses of the
experimernts. The authors appreciate Tohru Asami, Hiroshi Esaki and
Tatsuya Yamshita for their constructive comments.
10. Informative References
[1] "Winny on Wikipedia", <http://en.wikipedia.org/wiki/Winny>.
[2] Hiroshi Esaki, "The State of Traffic and the Effects of P2P",
Special Symposium on Broadband, September 2008 (in Japanese).
[3] Yoichi Yamazaki, "ISPs have Begun to Explore Tomorrow due to the
Expansion of Traffic", Nikkei Communications, December 2007 (in
Japanese).
[4] TVBank, "Live Delivery using `BB Broadcast'Achieving 96% Saving
in Traffic!", http:.wwww.tv-bank.com/jp/20081031.html, 2008 (in
Japanese).
[5] Open P4P, "P4P Field Tests: Yale-Pando-Verizon",
http://www.openp4p.net/front/fieldests, 2009.
[6] Ministry of Internal Affairs and Communications, "Disclosure of
the Report `Working Group on P2P Networks'",
http://www.soumu.go.jp/menu_news/s-news/2007/070629_11.html,
2007 (in Japanese).
[7] The Foundation for MultiMedia Communications, "The P2P Network
Experiment Council", http://www.fmmc.or.jp/P2P/about.htm, 2007
Kamei, et al. Expires May 24, 2011 [Page 14]
Internet-Draft P2P Experiments Japan November 2010
(in Japanese).
Authors' Addresses
Satoshi Kamei
NTT Service Integration Laboratories
3-9-11, Midori-cho
Musashino-shi, Tokyo 180-8585
JP
Phone: +81-422-59-6942
Email: kamei.satoshi@lab.ntt.co.jp
Tsuyoshi Momose
Cisco Systems G.K.
2-1-1 Nishi-Shinjuku
Shinjuku-ku, Tokyo 163-0409
JP
Phone: +81-3-5324-4154
Email: tmomose@cisco.com
Takeshi Inoue
NTT Communications
3-4-1, Shibaura
Minato-ku, Tokyo 108-8118
JP
Phone: +81-3-6733-7177
Email: inoue@jp.ntt.net
Tomohiro Nishitani
NTT Communications
1-2-20, Shibaura
Minato-ku, Tokyo 108-8118
JP
Phone: +81-50-3812-4742
Email: tomohiro.nishitani@ntt.com
Kamei, et al. Expires May 24, 2011 [Page 15]
