| DECADE | H. Song |
| Internet-Draft | N. Zong |
| Intended status: Informational | Huawei |
| Expires: November 06, 2012 | Y. Yang |
| Yale University | |
| R. Alimi | |
| May 7, 2012 |
DECoupled Application Data Enroute (DECADE) Problem Statement
draft-ietf-decade-problem-statement-06
Peer-to-peer (P2P) applications have become widely used on the Internet today and make up a large portion of the traffic in many networks. In P2P applications, one technique for reducing the transit and uplink P2P traffic is to introduce storage capabilities within the network. Traditional caches (e.g., P2P and Web caches) provide such storage, but they are complex (due to explicitly supporting individual P2P application protocols and cache refresh mechanisms) and they do not allow users to manage access to content in the cache. For example, content providers wishing to use in-network storage cannot easily control cache access and resource usage policies to satisfy their own requirements. This document discusses the introduction of in-network storage for P2P applications, and shows the need for a standard protocol for accessing this storage.
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 November 06, 2012.
Copyright (c) 2012 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.
Peer-to-peer (P2P) applications, including both P2P streaming and P2P filesharing applications, make up a large fraction of the traffic in many ISP networks today. One way to reduce bandwidth usage by P2P applications is to introduce storage capabilities in networks. Allowing P2P applications to store and retrieve data from inside networks can reduce traffic on the last-mile uplink, as well as on backbone and transit links.
P2P caches provide in-network storage and have been deployed in some networks. However, the current P2P caching architecture poses challenges to both P2P cache vendors and P2P application developers. For P2P cache vendors, it is challenging to support a number of continuously evolving P2P application protocols, due to lack of documentation, ongoing protocol changes, and rapid introduction of new features by P2P applications. For P2P applications, closed P2P caching systems limit P2P applications from effectively utilizing in-network storage. In particular, P2P caches typically do not allow users to explicitly store content into in-network storage. They also do not allow users to implement control over the content that has been placed into the in-network storage.
P2P applications suffer decreased efficiency, and the network infrastructure suffers increased load because there is no standardized interface for accessing storage and data transport services in the Internet.
Both of these challenges can be effectively addressed by using an open, standard protocol to access in-network storage [Data_Lockers]. P2P applications can store and retrieve content in the in-network storage, as well as control resources (e.g., bandwidth, connections) consumed by peers in a P2P application. As a simple example, a peer of a P2P application may upload to other peers through its in-network storage, saving its usage of last-mile uplink bandwidth.
In this document, we distinguish between two functional components of the native P2P application protocol: signaling and data access. Signaling includes operations such as handshaking and discovering peer and content availability. The data access component transfers content from one peer to another.
In essence, coupling of the signaling and data access makes in-network storage very complex to support various application services. However, these applications have common requirements for data access, making it possible to develop a standard protocol.
The following terms have special meaning in the definition of the in-network storage system.
The emergence of peer-to-peer (P2P) as a major network application (especially P2P file sharing and streaming) has led to substantial opportunities. The P2P paradigm can be utilized to design highly scalable and robust applications at low cost, compared to the traditional client-server paradigm.
However, P2P applications also face substantial design challenges. A particular problem facing P2P applications is the additional stress that they place on the network infrastructure. Furthermore, lack of infrastructure support can lead to unstable P2P application performance during peer churns and flash crowds, when a large group of users begin to retrieve the content during a short period of time. A potential way to solve it would be to make it possible for peers that were on bandwidth constrained access to put things in a place that is both not bandwidth constrained and accessible by other peers. These problems are now discussed in further detail.
A particular problem of the P2P paradigm is the stress that P2P application traffic places on the infrastructure of Internet service providers (ISPs). Multiple measurements (e.g., [Internet_Study_2008-2009]) have shown that P2P traffic has become a major type of traffic on some networks. Furthermore, the inefficiency of network-agnostic peering (at the P2P transmission level) leads to unnecessary traversal across network domains or spanning the backbone of a network [RFC5693].
Using network information alone to construct more efficient P2P swarms is not sufficient to reduce P2P traffic in access networks, as the total access upload traffic is equal to the total access download traffic in a traditional P2P system. On the other hand, it is reported that P2P traffic is becoming the dominant traffic on the access networks of some networks, reaching as high as 50-60% on the downlinks and 60-90% on the uplinks ([DCIA], [ICNP], [ipoque.P2P_survey.], [P2P_file_sharing]). Consequently, it becomes increasingly important to reduce upload access traffic, in addition to cross-domain and backbone traffic.
The inefficiency is also represented when traffic is sent upstream as many times as there are remote peers interested in getting the corresponding information. For example, the P2P application transfer completion times remain affected by potentially (relatively) slow upstream transmission. Similarly, the performance of real-time P2P applications may be affected by potentially (relatively) higher upstream latencies.
An effective technique to reduce P2P infrastructural stress and inefficiency is to introduce in-network storage. The existing in-network storage systems can be found in [RFC6392].
In the current Internet, in-network storage is introduced as P2P caches, either transparently or explicitly as a P2P peer. To provide service to a specific P2P application, the P2P cache server must support the specific signaling and transport protocols of the specific P2P application. This can lead to substantial complexity for the P2P Cache vendor.
First, there are many P2P applications on the Internet (e.g., BitTorrent, eMule, Flashget, and Thunder for file sharing; Abacast, Kontiki, Octoshape, PPLive, PPStream, and UUSee for P2P streaming). Consequently, a P2P cache vendor faces the challenge of supporting a large number of P2P application protocols, leading to product complexity and increased development cost.
Furthermore, a specific P2P application protocol may evolve continuously, to add new features or fix bugs. This forces a P2P cache vendor to continuously update to track the changes of the P2P application, leading to product complexity and increased costs.
Third, many P2P applications use proprietary protocols or support end-to-end encryption. This can render P2P caches ineffective. So these three problems make the P2P cache as a network middle-box, hard to support these P2P application distribution in their own ways.
Finally, a P2P cache is likely to be much better connected to end hosts than remote peers that connected to end hosts. Without the ability to manage bandwidth usage, the P2P cache may increase the volume of download traffic, which runs counter to the reduction of upload access traffic.
As P2P applications evolve, it has become increasingly clear that usage of in-network resources can improve user experience. For example, multiple P2P streaming systems seek additional in-network resources during a flash crowd, such as just before a major live streaming event. In asymmetric networks when the aggregated upload bandwidth of a channel cannot meet the download demand, a P2P application may seek additional in-network resources to maintain a stable system.
However, some P2P applications using in-network infrastructural resources require flexibility in implementing resource allocation policies. A major competitive advantage of many successful P2P systems is their substantial expertise in how to most efficiently utilize peer and infrastructural resources. For example, many live P2P systems have specific algorithms to select those peers that behave as stable, higher-bandwidth sources. Similarly, the higher-bandwidth sources frequently use algorithms to chose to which peers the source should send content. Developers of these systems continue to fine-tune these algorithms over time.
To permit developers to evolve and fine-tune their algorithms and policies, the in-network storage should expose basic mechanisms and allow as much flexibility as possible to P2P applications. This conforms to the end-to-end systems principle and allows innovation and satisfaction of specific business goals. Existing techniques for P2P application in-network storage lack these capabilities.
Usage scenarios are presented to illustrate the problems in both Content Distribution Network (CDN) and P2P scenarios.
When a BitTorrent client A uploads a block to multiple peers, the block traverses the last-mile uplink once for each peer. And after that, the peer B who just received the block from A also needs to upload through its own last-mile uplink to others when sharing this block. This is not an efficient use of the last-mile uplink. With in-network storage server however, the BitTorrent client may upload the block to its in-network storage. Peers may retrieve the block from the in-network storage, reducing the amount of data on the last-mile uplink. If supported by the in-network storage, a peer can also save the block in its own in-network storage while it is being retrieved; the block can then be uploaded from the in-network storage to other peers.
As previously discussed, BitTorrent or other P2P applications currently cannot explicitly manage which content is placed in the existing P2P caches, nor access and resource control polices. Applications need to retain flexibility to control the content distribution policies and topology among peers.
Content publishers may also utilize in-network storage. For example, consider a P2P live streaming application. A Content Publisher typically maintains a small number of sources, each of which distributes blocks in the current play buffer to a set of the P2P peers.
Some content publishers use another hybrid content distribution approach incorporating both P2P and CDN modes. As an example, Internet TV may be implemented as a hybrid CDN/P2P application by distributing content from central servers via a CDN, and also incorporating a P2P mode amongst endhosts and set-top boxes. In-network storage may be beneficial to hybrid CDN/P2P applications as well to support P2P distribution and to enable content publisher standard interfaces and controls.
However, there is no standard interface for different content publishers to access in-network storage. One streaming content publisher may need the existing in-network storage to support streaming signaling or such capability, such as transcoding capability, bitmap information, intelligent retransmission, etc, while a different content publisher may only need the in-network storage to distribute files. However it is reasonable that the application services are only supported by content publisher's original servers and clients, and intelligent data plane transport for those content publishers are supported by in-network storage.
A content publisher also benefits from a standard interface to access in-network storage servers provided by different providers. The standard interface must allow the content publisher to retain control over content placed in their own in-network storage, and grant access and resources only to the desired endhosts and peers.
In the hybrid CDN/P2P scenario, if only the endhosts can store content in the in-network storage server, the content must be downloaded and then uploaded over the last-mile access link before another peer may retrieve it from a in-network storage server. Thus, in this deployment scenario, it may be advantageous for a content publisher or CDN provider to store content in in-network storage servers.
There are several security considerations to the in-network storage.
An attacker can try to consume a large portion of the in-network storage, or exhaust the connections of the in-network storage through a Denial of Service (DoS) attack. Authentication, authorization and accounting mechanisms should be considered in the cross domain environment. Limitation of access from an administrative domain sets up barriers for content distribution.
Copyright and other laws may prevent the distribution of certain content in various localities. In-network storage operators may adopt system-wide ingress or egress filters to implement necessary policies for storing or retrieving content, and applications may apply DRM to the data stored in the network storage. However, the specification and implementation of such policies (e.g., filtering and DRM) is not in scope for the problem this document proposes solving.
If the content is stored in the provider-based in-network storage, there may be a privacy risk that the provider can correlate the people who are accessing the same data object using the same object identity. This correlation can be used to presume that they have the same interest, so as to use it as a basis for a phishing attack.
The modification threat is the danger that some unauthorized entity may alter in-transit in-network storage access messages generated on behalf of an authorized principal in such a way as to effect unauthorized management operations, including falsifying the value of an object. This threat may result in false data being supplied either through the data on a legitimate store being modified, or through a bogus store being introduced into the network.
A type of threat action whereby an unauthorized entity gains access to a system or performs a malicious act by illegitimately posing as an authorized entity. In the context of this spec, when accessing in-network storage, one malicious end host can try to act as another authorized end host or application server to access a protected resource in the in-network storage.
The disclosure threat is the danger of eavesdropping on the exchanges between in-network storage and application clients. Protecting against this threat may be required as a matter of application policy.
The message stream modification threat is the danger that messages may be maliciously re-ordered, delayed or replayed to an extent which is greater than can occur through natural network system, in order to effect unauthorized management operations to in-network storage. If the middle box, such like NAT (network address translator) or proxy between an end host and in-network storage is compromised, it is easy to do the stream modification attack.
There are no IANA considerations in this document.
We would like to thank the following people for contributing to this document:
David Bryan
Ronald Bonica
Kar Ann Chew
Roni Even
Lars Eggert
Francois Le Faucheur
Adrian Farrel
Yingjie Gu
David Harrington
Leif Johansson
Hongqiang Liu
Tao Ma
Borje Ohlman
Akbar Rahman
Robert Sparks
Peter Saint-Andre
Sean Turner
Yu-shun Wang
Richard Woundy
Yunfei Zhang
(To the RFC editor: Please remove this section and the related references in this section upon publication. The purpose of this section is to give the IESG and RFC editor a better understanding of the current P2P related work in IETF and the relationship with DECADE WG.)
Note that DECADE WG's work is independent of current IETF work on P2P. The ALTO work is aimed for better peer selection and the RELOAD [I-D.ietf-p2psip-base] protocol is used for P2P overlay maintenance and resource discovery.
The Peer to Peer Streaming Protocol effort in the IETF is investigating the specification of signaling protocols (called the PPSP tracker protocol and peer protocol) for multiple entities (e.g., intelligent endpoints, caches, content distribution network nodes, and/or other edge devices) to participate in P2P streaming systems in both fixed and mobile Internet. As discussed in the PPSP problem statement, one important PPSP use case is the support of an in-network edge cache for P2P Streaming. However, this approach to providing in-network cache has different applicability, different objectives and different implications for the in-network cache operator. The goal of DECADE WG is to provide in-network storage service that can be used for any application transparently to the in-network storage operator: it can be used for any P2P Streaming application (whether it supports PPSP protocols or not), for any other P2P application, and for non P2P applications that simply want to benefit from in-network storage. With DECADE, the operator is providing a generic in-network storage service that can be used by any application without application involvement or awareness by the operator; in the PPSP cache use case, the cache operator is participating in the specific P2P streaming service.
DECADE and PPSP can both contribute independently, and (where appropriate) simultaneously, to making content available closer to peers. Here are a number of example scenarios:
An ALTO (Application Layer Traffic Optimization) server provides P2P applications with network information so that they can perform better-than-random initial peer selection [RFC5693]. However, there are limitations on what ALTO can achieve alone. For example, network information alone cannot reduce P2P traffic in access networks, as the total access upload traffic is equal to the total access download traffic in a traditional P2P system. Consequently, it becomes increasingly important to complement the ALTO effort and reduce upload access traffic, in addition to cross-domain and backbone traffic.
The IETF Low Extra Delay Background Transport (LEDBAT) Working Group is focusing on techniques that allow large amounts of data to be consistently transmitted without substantially affecting the delays experienced by other users and applications. It is expected that some P2P applications would start using such techniques, thereby somewhat alleviating the perceivable impact (at least on other applications) of their high volume traffic. However, such techniques may not be adopted by all P2P applications. Also, when adopted, these techniques do not remove all inefficiencies, such as those associated with traffic being sent upstream as many times as there are remote peers interested in getting the corresponding information. For example, the P2P application transfer completion times remain affected by potentially (relatively) slow upstream transmission. Similarly, the performance of real-time P2P applications may be affected by potentially (relatively) higher upstream latencies.