RFC 7846
Peer-to-Peer Streaming Tracker Protocol (PPSTP)
Pages: 55
Proposed Standard
Proposed Standard
Part 1 of 3 – Pages 1 to 16
Internet Engineering Task Force (IETF) R. Cruz Request for Comments: 7846 M. Nunes Category: Standards Track IST/INESC-ID/INOV ISSN: 2070-1721 J. Xia R. Huang, Ed. Huawei J. Taveira IST/INOV D. Lingli China Mobile May 2016 Peer-to-Peer Streaming Tracker Protocol (PPSTP)Abstract
This document specifies the base Peer-to-Peer Streaming Tracker Protocol (PPSTP) version 1, an application-layer control (signaling) protocol for the exchange of meta information between trackers and peers. The specification outlines the architecture of the protocol and its functionality; it also describes message flows, message processing instructions, message formats, formal syntax, and semantics. The PPSTP enables cooperating peers to form content- streaming overlay networks to support near real-time delivery of structured media content (audio, video, and associated timed text and metadata), such as adaptive multi-rate, layered (scalable), and multi-view (3D) videos in live, time-shifted, and on-demand modes. Status of This Memo This is an Internet Standards Track document. This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Further information on Internet Standards is available in Section 2 of RFC 7841. Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at http://www.rfc-editor.org/info/rfc7846.
Copyright Notice Copyright (c) 2016 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.Table of Contents
1. Introduction ....................................................4 1.1. Terminology ................................................4 1.2. Design Overview ............................................6 1.2.1. Typical PPSP Session ................................7 1.2.2. Example of a PPSP Session ...........................7 2. Protocol Architecture and Functional View ......................10 2.1. Messaging Model ...........................................10 2.2. Request/Response Model ....................................10 2.3. State Machines and Flows of the Protocol ..................12 2.3.1. Normal Operation ...................................14 2.3.2. Error Conditions ...................................15 3. Protocol Specification .........................................16 3.1. Presentation Language .....................................16 3.2. Resource Element Types ....................................16 3.2.1. Version ............................................16 3.2.2. Peer Number Element ................................17 3.2.3. Swarm Action Element ...............................18 3.2.4. Peer Information Elements ..........................18 3.2.5. Statistics and Status Information Element ..........20 3.3. Requests and Responses ....................................21 3.3.1. Request Types ......................................21 3.3.2. Response Types .....................................21 3.3.3. Request Element ....................................22 3.3.4. Response Element ...................................23 3.4. PPSTP Message Element .....................................24 4. Protocol Specification: Encoding and Operation .................24 4.1. Requests and Responses ....................................25 4.1.1. CONNECT Request ....................................25 4.1.1.1. Example ...................................28 4.1.2. FIND Request .......................................32 4.1.2.1. Example ...................................33
4.1.3. STAT_REPORT Request ................................34
4.1.3.1. Example ...................................35
4.2. Response Element in Response Messages .....................36
4.3. Error and Recovery Conditions .............................37
4.4. Parsing of Unknown Fields in message-body .................38
5. Operations and Manageability ...................................38
5.1. Operational Considerations ................................38
5.1.1. Installation and Initial Setup .....................38
5.1.2. Migration Path .....................................39
5.1.3. Requirements on Other Protocols and
Functional Components ..............................39
5.1.4. Impact on Network Operation ........................39
5.1.5. Verifying Correct Operation ........................40
5.2. Management Considerations .................................40
5.2.1. Interoperability ...................................40
5.2.2. Management Information .............................40
5.2.3. Fault Management ...................................41
5.2.4. Configuration Management ...........................41
5.2.5. Accounting Management ..............................41
5.2.6. Performance Management .............................41
5.2.7. Security Management ................................41
6. Security Considerations ........................................42
6.1. Authentication between Tracker and Peers ..................42
6.2. Content Integrity Protection against Polluting
Peers/Trackers ............................................43
6.3. Residual Attacks and Mitigation ...........................43
6.4. Pro-incentive Parameter Trustfulness ......................44
6.5. Privacy for Peers .........................................44
7. Guidelines for Extending PPSTP .................................45
7.1. Forms of PPSTP Extension ..................................45
7.2. Issues to Be Addressed in PPSTP Extensions ................47
8. IANA Considerations ............................................48
8.1. MIME Type Registry ........................................48
8.2. PPSTP Version Number Registry .............................49
8.3. PPSTP Request Type Registry ...............................49
8.4. PPSTP Error Code Registry .................................50
9. References .....................................................51
9.1. Normative References ......................................51
9.2. Informative References ....................................53
Acknowledgments ...................................................54
Authors' Addresses ................................................55
1. Introduction
The Peer-to-Peer Streaming Protocol (PPSP) is composed of two protocols: the Tracker Protocol (defined in this document) and the Peer Protocol (defined in [RFC7574]). [RFC6972] specifies that the Tracker Protocol should standardize the messages between PPSP peers and PPSP trackers and also defines the requirements. The Peer-to-Peer Streaming Tracker Protocol (PPSTP) provides communication between trackers and peers by which peers send meta information to trackers, report streaming status, and obtain peer lists from trackers. The PPSP architecture requires PPSP peers to be able to communicate with a tracker in order to participate in a particular streaming content swarm. This centralized tracker service is used by PPSP peers for acquisition of peer lists. The signaling and the media data transfer between PPSP peers is not in the scope of this specification. This document introduces a base Peer-to-Peer Streaming Tracker Protocol (PPSTP) that satisfies the requirements in [RFC6972].1.1. Terminology
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119]. absolute time: Expressed as ISO 8601 timestamps, using zero UTC offset. Fractions of a second may be indicated, for example, December 25, 2010 at 14h56 and 20.25 seconds in basic format is 20101225T145620.25Z and in extended format is 2010-12-25T14:56:20.25Z. chunk: An uniformly atomic subset of the resource that constitutes the basic unit of data organized in P2P streaming for storage, scheduling, advertisement, and exchange among peers. chunk ID: A unique resource identifier for a chunk. The identifier type depends on the addressing scheme used, i.e., an integer, an HTTP-URL, and possibly a byte-range. The identifier type is described in the Media Presentation Description (MPD). LEECH: The peers in a swarm that download content from other peers as well as contribute downloaded content with others. A LEECH should join the swarm with uncompleted media content.
MPD (Media Presentation Description): Formalized description for a
media presentation, i.e., describes the structure of the media,
namely, the representations, the codecs used, the chunks, and the
corresponding addressing scheme.
peer: A participant in a P2P streaming system that not only receives
streaming content, but also caches and streams streaming content
to other participants.
peer ID: The identifier of a peer such that other peers, or the
Tracker, can refer to the peer using its ID. The peer ID is
mandatory, can take the form of a universally unique identifier
(UUID), defined in [RFC4122], and can be bound to a network
address of the peer, i.e., an IP address or a uniform resource
identifier/locator (URI/URL) that uniquely identifies the
corresponding peer in the network. The peer ID and any required
security certificates are obtained from an offline enrollment
server.
peer list: A list of peers that are in the same swarm maintained by
the tracker. A peer can fetch the peer list of a swarm from the
tracker.
PPSP: The abbreviation of Peer-to-Peer Streaming Protocol.
PPSTP: The abbreviation of Peer-to-Peer Streaming Tracker Protocol.
SEEDER: The peers in a swarm that only contribute the content they
have to others. A SEEDER should join the swarm with complete
media content.
service portal: A logical entity typically used for client enrollment
and for publishing, searching, and retrieving content information.
It is usually located in a server of a content provider.
swarm: A group of peers that exchange data to distribute chunks of
the same content (e.g., video/audio program, digital file, etc.)
at a given time.
swarm ID: The identifier of a swarm containing a group of peers
sharing common streaming content. The swarm ID may use a
universally unique identifier (UUID), e.g., a 64- or 128-bit datum
to refer to the content resource being shared among peers.
tracker: A directory service that maintains a list of peers
participating in a specific audio/video channel or in the
distribution of a streaming file. It is a logical component that
can be deployed in a centralized or distributed way.
transaction ID: The identifier of a request from the peer to the
tracker. It is used to disambiguate responses that may arrive in
a different order than the corresponding requests.
1.2. Design Overview
The functional entities related to peer-to-peer streaming protocols
are the Client Media Player, the service portal, the tracker, and the
peers. The complete description of Client Media Player and service
portal is not discussed here, as they are not in the scope of the
specification. The functional entities directly involved in PPSTP
are trackers and peers (which may support different capabilities).
The Client Media Player is a logical entity providing direct
interface to the end user at the client device and includes the
functions to select, request, decode, and render content. The Client
Media Player may interface with the local peer application using the
standard format for HTTP request and response messages [RFC7230].
The service portal is a logical entity typically used for client
enrollment and for publishing, searching, and retrieving content
information.
A peer corresponds to a logical entity (typically in a user device)
that actually participates in sharing media content. Peers are
organized in various swarms; each swarm corresponds to the group of
peers streaming certain content at any given time.
A tracker is a logical entity that maintains the lists of peers
storing chunks for a specific live media channel or on-demand media
streaming content, answers queries from peers, and collects
information on the activity of peers. While a tracker may have an
underlying implementation consisting of more than one physical node,
logically, the tracker can most simply be thought of as a single
element; in this document, it will be treated as a single logical
entity. Communication between these physical nodes to present them
as a single tracker to peers is not considered in PPSTP, which is a
protocol between a tracker and a peer.
PPSTP is not used to exchange actual content data (either on demand
or live streaming) with peers, but information about which peers can
provide the content. PPSTP is not designed for applications for
which in-sync reception is needed.
1.2.1. Typical PPSP Session
When a peer wants to receive streaming of selected content (LEECH mode): 1. Peer connects to a tracker and joins a swarm. 2. Peer acquires a list of other peers in the swarm from the tracker. 3. Peer exchanges its content availability with the peers on the obtained peer list. 4. Peer identifies the peers with desired content. 5. Peer requests content from the identified peers. When a peer wants to share streaming content (SEEDER mode) with other peers: 1. Peer connects to a tracker. 2. Peer sends information to the tracker about the swarms it belongs to (joined swarms). 3. Peer waits for other peers in LEECH mode to connect with it (see steps 3-5 in the previous list). After having been disconnected due to some termination conditions or user controls, a peer can resume previous activity by connecting and re-joining the corresponding swarm(s).1.2.2. Example of a PPSP Session
In order to be able to bootstrap in the P2P network, a peer must first obtain a peer ID and any required security certificates or authorization tokens from an enrollment service (end-user registration). The peer ID MUST be unique (see the definition of "peer ID" in Section 1.1); however, the representation of the peer ID is not considered in this document.
+--------+ +--------+ +--------+ +---------+ +--------+ | Player | | Peer_1 | | Portal | | Tracker | | Peer_2 | +--------+ +--------+ +--------+ +---------+ +--------+ | | | | | (a) |--Page request----------------->| | | |<--------------Page with links--| | | |--Select stream (MPD request)-->| | | |<--------------------OK+MPD(x)--| | | (b) |--Start/Resume->|--CONNECT(join x)------------>| | |<-----------OK--|<----------------OK+Peerlist--| | | | | | |--Get(chunk)--->|<---------- (Peer protocol) ------------->| |<--------chunk--|<---------------------------------chunks--| : : : : : | |--STAT_REPORT---------------->| | | |<-------------------------OK--| | : : : : : | |--FIND----------------------->| | | |<----------------OK+Peerlist--| | : : : : : |--Get(chunk)--->|<---------- (Peer protocol) ------------->| |<--------chunk--|<---------------------------------chunks--| : : : : : Figure 1: A Typical PPSP Session for Streaming Content To join an existing P2P streaming service and to participate in content sharing, a peer must first locate a tracker. As illustrated in Figure 1, a P2P streaming session may be initiated starting at point (a), with the Client Media Player browsing for the desired content in order to request it (to the local Peer_1 in the figure), or resume a previously initiated stream, but starting at point (b). For this example, the Peer_1 is in mode LEECH. At point (a) in Figure 1, the Client Media Player accesses the portal and selects the content of interest. The portal returns the Media Presentation Description (MPD) file that includes information about the address of one or more trackers (which can be grouped by tiers of priority) that control the swarm x for that media content (e.g., content x). With the information from the MPD, the Client Media Player is able to trigger the start of the streaming session, requesting to the local Peer_1 the chunks of interest.
The PPSP streaming session is then started (or resumed) at Peer_1 by sending a PPSTP CONNECT message to the tracker in order to join swarm x. The tracker will then return the OK response message containing a peer list, if the CONNECT message is successfully accepted. From that point, every chunk request is addressed by Peer_1 to its neighbors (Peer_2 in Figure 1) using a peer protocol, e.g., [RFC7574], returning the received chunks to the Client Media Player. Once connected, Peer_1 needs to periodically report its status and statistics data to the tracker using a STAT_REPORT message. If Peer_1 needs to refresh its neighborhood (for example, due to churn), it will send a PPSTP FIND message (with the desired scope) to the tracker. Peers that are only SEEDERs (i.e., serving content to other peers), as are the typical cases of service provider P2P edge caches and/or media servers, trigger their P2P streaming sessions for content x, y, z... (Figure 2), not from Media Player signals, but from some "Start" activation signal received from the service provider provisioning mechanism. In this particular case, the peer starts or resumes all its streaming sessions just by sending a PPSTP CONNECT message to the tracker (Figure 2), in order to "join" all the requested swarms. Periodically, the peer also reports its status and statistics data to the tracker using a PPSTP STAT_REPORT message. +---------+ +---------+ | SEEDER | | Tracker | +---------+ +---------+ | | Start->|--CONNECT (join x,y,z)-------->| |<--------------------------OK--| : : | | |--STAT_REPORT----------------->| |<--------------------------Ok--| : : | | |--STAT_REPORT----------------->| |<--------------------------Ok--| : : Figure 2: A Typical PPSP Session for a Streaming SEEDER
The specification of the mechanisms used by the Client Media Player (or provisioning process) and the peer to signal start/resume of streams, request media chunks, and obtain a peer ID, security certificates, or tokens is not in the scope of this document.2. Protocol Architecture and Functional View
PPSTP is designed with a layered approach i.e., a PPSTP Request/Response layer, a Message layer, and a Transport layer (see Figure 3). +------------------------+ | Application | +------------------------+ |(PPSTP) Request/Response| |------------------------| | (HTTP) Message | +------------------------+ | Transport | +------------------------+ Figure 3: Abstract Layering of PPSTP The PPSTP Request/Response layer deals with the interactions between tracker and peers using request and response messages. The Message layer deals with the framing format for encoding and transmitting data through the underlying transport protocol, as well as the asynchronous nature of the interactions between tracker and peers. The Transport layer is responsible for the actual transmission of requests and responses over network transports, including the determination of the connection to use for a request or response message when using TCP or Transport Layer Security (TLS) [RFC5246] over it.2.1. Messaging Model
The messaging model of PPSTP aligns with HTTP, which is currently in version 1.1 [RFC7230], and the semantics of its messages. PPSTP is intended to also support future versions of HTTP.2.2. Request/Response Model
PPSTP uses a design like REST (Representational State Transfer) with the goal of leveraging current HTTP implementations and infrastructure, as well as familiarity with existing REST-like
services in popular use. PPSTP messages use the UTF-8 character set [RFC3629] and are either requests from peers to a tracker service or responses from a tracker service to peers. The request and response semantics are carried as entities (header and body) in messages that correspond to either HTTP request methods or HTTP response codes, respectively. PPSTP uses the HTTP POST method to send parameters in requests. PPSTP messages use JavaScript Object Notation (JSON) [RFC7159] to encode message bodies. Peers send requests to trackers. Trackers send a single response for each request though both requests and responses can be subject to fragmentation of messages in transport. The request messages of the base protocol are listed in Table 1: +------------------------------+ | PPSTP Request Messages | +------------------------------+ | CONNECT | | FIND | | STAT_REPORT | +------------------------------+ Table 1: Request Messages CONNECT: This request message is used when a peer registers in the tracker to notify it about participation in the named swarm(s). If the peer is already registered in the tracker, this request message simply notifies the tracker about participation in the named swarm(s). The tracker records the peer ID, connect-time (referenced to the absolute time), peer IP addresses (and associated location information), link status, and peer mode for the named swarm(s). The tracker also changes the content availability of the valid named swarm(s), i.e., changes the peer's lists of the corresponding swarm(s) for the requesting peer ID. On receiving a CONNECT message, the tracker first checks the peer mode type (SEEDER/LEECH) for the specified swarm(s) and then decides the next steps (see Section 4.1 for more details). FIND: This request message is used by peers to request a list of peers active in the named swarm from the tracker whenever needed. On receiving a FIND message, the tracker finds the peers listed in the content status of the specified swarm that can satisfy the requesting peer's requirements and returns the list to the
requesting peer. To create the peer list, the tracker may take
peer status, capabilities, and peer priority into consideration.
Peer priority may be determined by network topology preference,
operator policy preference, etc.
STAT_REPORT:
This request message is used to allow an active peer to send
status (and optionally statistic data) to the tracker to signal
continuing activity. This request message MUST be sent
periodically to the tracker while the peer is active in the
system.
2.3. State Machines and Flows of the Protocol
The state machine for the tracker is very simple, as shown in Figure
4. Peer ID registrations represent a dynamic piece of state
maintained by the network.
--------------------------------------------
/ \
| +------------+ +=========+ +======+ |
\-| TERMINATED |<---| STARTED |<---| INIT |<-/
+------------+ +=========+ +======+
(Transient) \- (start tracker)
Figure 4: Tracker State Machine
When there are no peers connected in the tracker, the state machine
is in INIT state.
When the first peer connects to register with its peer ID, the state
machine moves from INIT to STARTED. As long as there is at least one
active registration of a peer ID, the state machine remains in
STARTED state. When the last peer ID is removed, the state machine
transitions to TERMINATED. From there, it immediately transitions
back to INIT state. Because of this, TERMINATED state is transient.
Once in STARTED state, each peer is instantiated (per peer ID) in the
tracker state machine with a dedicated transaction state machine
(Figure 5), which is deleted when the peer ID is removed.
-------------------------------------------- / \ | +------------+ +=========+ +======+ | \-| TERMINATED |<---| STARTED |<---| INIT |<-/ +------------+ +=========+ +======+ (Transient) | (1) \- (start tracker) V +-----------+ +-------+ rcv CONNECT (Transient) | TERMINATE | | START | --------------- (1) +-----------+ +-------+ strt init timer rcv FIND (B) ^ | rcv STAT_REPORT (B) | | on registration error (B)| v on action error (A) | +------------+ ---------------- +<--| PEER | (Transient) stop init timer | | REGISTERED | snd error | +------------+ | | on timeout (D) | | process swarm actions ---------------- | | --------------------- (2) stop track timer | | snd OK (PeerList) clean peer info | / stop init timer del registration | / strt track timer | / | | | | rcv FIND STAT_REPORT ERR(C) \ | ---- --------------- (3) FIND ERR(C) ---- \ | / \ snd OK (PeerList) CONNECT ERR(C) / \ | | | | rst track timer rcv CONNECT | (4) | | | | | ----------- | v | v v | rcv STAT_REPORT snd OK \ +==============+ / --------------- (3) rst track timer ----| TRACKING |---- snd OK response snd error (C) +==============+ rst track timer Figure 5: "Per-Peer-ID" State Machine and Flow Diagram Unlike the tracker state machine, which exists even when no peer IDs are registered, the "per-Peer-ID" State Machine is instantiated only when the peer ID starts registration in the tracker and is deleted when the peer ID is de-registered/removed. This allows for an implementation optimization whereby the tracker can destroy the objects associated with the "per-Peer-ID" State Machine once it enters the TERMINATE state (Figure 5). When a new peer ID is added, the corresponding "per-Peer-ID" State Machine is instantiated, and it moves into the PEER REGISTERED state. Because of that, the START state here is transient.
When the peer ID is no longer bound to a registration, the "per-Peer- ID" State Machine moves to the TERMINATE state, and the state machine is destroyed. During the lifetime of streaming activity of a peer, the instantiated "per-Peer-ID" State Machine progresses from one state to another in response to various events. The events that may potentially advance the state include: o Reception of CONNECT, FIND, and STAT_REPORT messages o Timeout events The state diagram in Figure 5 illustrates state changes, together with the causing events and resulting actions. Specific error conditions are not shown in the state diagram.2.3.1. Normal Operation
For normal operation, the process consists of the following steps: 1) When a peer wants to access the system, it needs to register with a tracker by sending a CONNECT message asking for the swarm(s) it wants to join. This request from a new peer ID triggers the instantiation in the tracker of a "per-Peer-ID" State Machine. In the START state of the new "per-Peer-ID" State Machine, the tracker registers the peer ID and associated information (IP addresses), starts the "init timer", and moves to PEER REGISTERED state. 2) In PEER REGISTERED state, if the peer ID is valid, the tracker either: a) processes the requested action(s) for the valid swarm information contained in the CONNECT requests, and if successful, the tracker stops the "init timer", starts the "track timer", and sends the response to the peer (the response may contain the appropriate list of peers for the joining swarm(s), as detailed in Section 4.1), or b) moves the valid FIND request to TRACKING state. 3) In TRACKING state, STAT_REPORT or FIND messages received from that peer ID will reset the "track timer", and the tracker responds to the requests with the following, respectively:
a) a successful condition, or
b) a successful condition containing the appropriate list of peers
for the named swarm (Section 4.2).
4) While in TRACKING state, a CONNECT message received from that peer
ID with valid swarm action information (Section 4.1.1) resets the
"track timer", and the tracker responds to the request with a
successful condition.
2.3.2. Error Conditions
Peers are required not to generate protocol elements that are
invalid. However, several situations may lead to abnormal conditions
in the interaction with the tracker. These situations may be related
to peer malfunction or communication errors. The tracker reacts to
these abnormal situations depending on its current state related to a
peer ID, as follows:
A) In PEER REGISTERED state, when a CONNECT request only contains
invalid swarm actions (Section 4.1.1), the tracker responds with a
PPSTP error code as specified in Section 4.3, deletes the
registration, and transitions to TERMINATE state for that peer ID.
The state machine is destroyed.
B) In PEER REGISTERED state, if the peer ID is considered invalid (in
the case of a CONNECT request or in the case of FIND or
STAT_REPORT requests received from an unregistered peer ID), the
tracker responds with either a 06 (Authentication Required)
error_code or a 03 (Forbidden Action) error_code as described in
Section 4.3 and transitions to TERMINATE state for that peer ID.
The state machine is destroyed.
C) In TRACKING state (while the "track timer" has not expired),
receiving a CONNECT message from a peer ID with invalid swarm
actions (Section 4.1.1) or receiving a FIND/STAT_REPORT message
from a peer ID with an invalid swarm ID is considered an error
condition. The tracker responds with the corresponding error code
(described in Section 4.3).
D) In TRACKING state, without receiving messages from the peer on
timeout (the "track timer" has expired), the tracker cleans all
the information associated with the peer ID in all swarms it was
joined, deletes the registration, and transitions to TERMINATE
state for that peer ID. The state machine is destroyed.
NOTE: These situations may correspond to malfunctions at the peer or to malicious conditions. As a preventive measure, the tracker proceeds to TERMINATE state for that peer ID.
(page 16 continued on part 2)