BFCPbis Working Group T. Kristensen, Ed.
Internet-Draft C. Eckel
Intended status: Standards Track A. Heggestad
Expires: August 20, 2012 G. Sandbakken
Cisco
February 17, 2012
Revision of the Binary Floor Control Protocol (BFCP) for use over an
unreliable transport
draft-ietf-bfcpbis-rfc4582bis-01
Abstract
This draft describes how to extend the Binary Floor Control Protocol
(BFCP) for use over an unreliable transport. It details the
differences from the BFCP protocol definition document and the
Session Description Protocol (SDP) format specified for BFCP streams.
Status of this Memo
This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF). Note that other groups may also distribute
working documents as Internet-Drafts. The list of current Internet-
Drafts is at http://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress."
This Internet-Draft will expire on August 20, 2012.
Copyright Notice
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
Kristensen, et al. Expires August 20, 2012 [Page 1]
Internet-Draft BFCP-Unreliable February 2012
the Trust Legal Provisions and are provided without warranty as
described in the Simplified BSD License.
Table of Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4
2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 4
3. Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.1. Alternatives Considered . . . . . . . . . . . . . . . . . 6
3.1.1. ICE TCP . . . . . . . . . . . . . . . . . . . . . . . 6
3.1.2. Teredo . . . . . . . . . . . . . . . . . . . . . . . . 6
3.1.3. GUT . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.1.4. UPnP IGD . . . . . . . . . . . . . . . . . . . . . . . 7
3.1.5. NAT PMP . . . . . . . . . . . . . . . . . . . . . . . 7
4. Difference from RFC4582 . . . . . . . . . . . . . . . . . . . 8
4.1. Overview of Operation (4) . . . . . . . . . . . . . . . . 8
4.1.1. Floor Participant to Floor Control Server
Interface (4.1) . . . . . . . . . . . . . . . . . . . 8
4.2. COMMON-HEADER Format (5.1) . . . . . . . . . . . . . . . . 8
4.3. ERROR-CODE (5.2.6) . . . . . . . . . . . . . . . . . . . . 10
4.4. FloorRequestStatusAck (5.3.14) . . . . . . . . . . . . . . 10
4.5. ErrorAck (5.3.15) . . . . . . . . . . . . . . . . . . . . 11
4.6. FloorStatusAck (5.3.16) . . . . . . . . . . . . . . . . . 11
4.7. Goodbye (5.3.17) . . . . . . . . . . . . . . . . . . . . . 12
4.8. GoodbyeAck (5.3.18) . . . . . . . . . . . . . . . . . . . 12
4.9. Transport (6) . . . . . . . . . . . . . . . . . . . . . . 12
4.9.1. Reliable Transport (6.1) . . . . . . . . . . . . . . . 13
4.9.2. Unreliable Transport (6.2) . . . . . . . . . . . . . . 14
4.9.2.1. Congestion Control . . . . . . . . . . . . . . . . 15
4.9.2.2. ICMP Error Handling . . . . . . . . . . . . . . . 15
4.9.3. Large Message Considerations . . . . . . . . . . . . . 16
4.9.3.1. Fragmentation Handling . . . . . . . . . . . . . . 16
4.10. Lower-Layer Security (7) . . . . . . . . . . . . . . . . . 16
4.11. Protocol Transactions (8) . . . . . . . . . . . . . . . . 17
4.12. Server Behavior (8.2) . . . . . . . . . . . . . . . . . . 17
4.13. Timers (8.3) . . . . . . . . . . . . . . . . . . . . . . . 18
4.14. Request Retransmission Timer, T1 (8.3.1) . . . . . . . . . 18
4.15. Response Retransmission Timer, T2 (8.3.2) . . . . . . . . 18
4.16. Timer Values (8.3.3) . . . . . . . . . . . . . . . . . . . 18
4.17. Authentication and Authorization (9) . . . . . . . . . . . 19
4.17.1. TLS Based Mutual Authentication (9.1) . . . . . . . . 19
4.18. Receiving a Response [to a FloorRequest Message]
(10.1.2) . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.19. Receiving a Response [to a FloorRelease Message]
(10.2.2) . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.20. Receiving a Response [to a ChairAction Message] (11.2) . . 20
4.21. Receiving a Response [to a FloorQuery Message] (12.1.2) . 20
Kristensen, et al. Expires August 20, 2012 [Page 2]
Internet-Draft BFCP-Unreliable February 2012
4.22. Receiving a Response [to a FloorRequestQuery Message]
(12.2.2) . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.23. Receiving a Response [to a UserQuery Message] (12.3.2) . . 20
4.24. Receiving a Response [to a Hello Message] (12.4.2) . . . . 20
4.25. Reception of a FloorRequestStatus Message (13.1.3) . . . . 21
4.26. Reception of a FloorStatus Message (13.5.3) . . . . . . . 21
4.27. Reception of an Error Message (13.8.1) . . . . . . . . . . 21
4.28. Security Considerations (14) . . . . . . . . . . . . . . . 21
4.29. IANA Considerations - Primitive Subregistry (15.2) . . . . 21
4.30. IANA Considerations - Error Code Subregistry (15.4) . . . 22
4.31. Example Call Flows for BFCP over Unreliable Transport
(Appendix A) . . . . . . . . . . . . . . . . . . . . . . . 22
5. Revision of RFC4583 . . . . . . . . . . . . . . . . . . . . . 25
5.1. Fields in the 'm' Line (3) . . . . . . . . . . . . . . . . 26
5.2. Authentication (8) . . . . . . . . . . . . . . . . . . . . 26
5.3. Security Considerations (10) . . . . . . . . . . . . . . . 26
5.4. Registration of SDP 'proto' Values (11.1) . . . . . . . . 26
6. NAT Traversal . . . . . . . . . . . . . . . . . . . . . . . . 27
7. Remaining Work . . . . . . . . . . . . . . . . . . . . . . . . 27
8. Contributing Authors . . . . . . . . . . . . . . . . . . . . . 28
9. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 28
10. References . . . . . . . . . . . . . . . . . . . . . . . . . . 28
10.1. Normative References . . . . . . . . . . . . . . . . . . . 28
10.2. Informative References . . . . . . . . . . . . . . . . . . 29
Appendix A. Change History . . . . . . . . . . . . . . . . . . . 30
A.1. draft-ietf-bfcpbis-rfc4582bis-00 to -01 . . . . . . . . . 30
A.2. draft-sandbakken-dispatch-bfcp-udp-03 to
draft-ietf-bfcpbis-rfc4582bis-00 . . . . . . . . . . . . . 30
A.3. draft-sandbakken-dispatch-bfcp-udp-02 to -03 . . . . . . . 31
A.4. draft-sandbakken-dispatch-bfcp-udp-01 to -02 . . . . . . . 31
A.5. draft-sandbakken-dispatch-bfcp-udp-00 to -01 . . . . . . . 31
A.6. draft-sandbakken-xcon-bfcp-udp-02 to
draft-sandbakken-dispatch-bfcp-udp-00 . . . . . . . . . . 32
A.7. draft-sandbakken-xcon-bfcp-udp-01 to -02 . . . . . . . . . 33
A.8. draft-sandbakken-xcon-bfcp-udp-00 to -01 . . . . . . . . . 33
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 33
Kristensen, et al. Expires August 20, 2012 [Page 3]
Internet-Draft BFCP-Unreliable February 2012
1. Introduction
This draft describes how to extend the BFCP protocol to support
unreliable transport. Minor changes to the transaction model are
introduced in that all requests now have an appropriate response to
complete the transaction. The requests are sent with a retransmit
timer associated with the response to achieve reliability.
This extension does not change the semantics of BFCP. It permits UDP
as an alternate transport. Existing implementations, in the spirit
of the approach detailed in earlier versions of this draft (see
Appendix A), have demonstrated the approach to be feasible. Initial
compatibility among implementations has been achieved at previous
interoperability events. The purpose of this draft is to formalize
and publish the extension from the standard specification to
facilitate complete interoperability between implementations.
The content of this draft relates to the BFCP protocol specification
[RFC4582] and the SDP format for describing BFCP streams [RFC4583].
This draft is written with the goal of identifying the extensions
associated with adding support for UDP as an alternate transport to
an existing BFCP implementation.
2. 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].
3. Motivation
In existing video conferencing deployments, BFCP is used to manage
the floor for the content sharing associated with the conference.
For peer to peer scenarios, including business to business
conferences and point to point conferences in general, it is
frequently the case that one or both endpoints exists behind a NAT/
firewall. BFCP roles are negotiated in the offer/answer exchange as
specified in [RFC4583], resulting in one endpoint being responsible
for opening the TCP connection used for the BFCP communication.
Kristensen, et al. Expires August 20, 2012 [Page 4]
Internet-Draft BFCP-Unreliable February 2012
+---------+
| Network |
+---------+
+-----+ / \ +-----+
| NAT |/ \| NAT |
+-----+ +-----+
+----+ / \ +----+
|BFCP|/ \|BFCP|
| UA | | UA |
+----+ +----+
Figure 1: Use Case
The communication session between the video conferencing endpoints
typically consists of a number of RTP over UDP media streams, for
audio and video, and a BFCP connection for floor control. Existing
deployments are most common in, but not limited to, enterprise
networks. In existing deployments, NAT/firewall traversal for the
RTP streams works using ICE and/or other methods, including those
described in [I-D.ietf-mmusic-media-path-middleboxes].
When enhancing an existing SIP based video conferencing deployment
with support for content sharing, the BFCP connection often poses a
problem. The reasons for this fall into two general classes. First,
there may be a strong preference for UDP based signaling in general.
On high capacity endpoints (e.g. PSTN gateways or SIP/H.323 inter-
working gateways), TCP can suffer from head of line blocking, and it
uses many kernel buffers. Network operators view UDP as a way to
avoid both of these. Second, establishment and traversal of the TCP
connection involving ephemeral ports, as is typically the case with
BFCP over TCP, can be problematic, as described in Appendix A of
[I-D.ietf-mmusic-ice-tcp]. A broad study of NAT behavior and peer-
to-peer TCP establishment for a comprehensive set of TCP NAT
traversal techniques over a wide range of commercial NAT products
concluded it was not possible to establish a TCP connection in 11% of
the cases [IMC05]. The results are worse when focusing on enterprise
NATs. A study of hole punching as a NAT traversal technique across a
wide variety of deployed NATs reported consistently higher success
rates when using UDP than when using TCP [P2PNAT].
To overcome the problems with establishing TCP flows between BFCP
entities, this draft defines UDP as an alternate transport for BFCP,
leveraging the same mechanisms in place for the RTP over UDP media
streams for the BFCP communication. When using UDP as the transport,
it is RECOMMENDED to follow the guidelines provided in [RFC5405].
NAT traversal for BFCP over UDP entities is discussed in more detail
in Section 6.
Kristensen, et al. Expires August 20, 2012 [Page 5]
Internet-Draft BFCP-Unreliable February 2012
The authors view this extension as a pragmatic solution to an
existing deployment challenge.
3.1. Alternatives Considered
In selecting the approach of defining UDP as an alternate transport
for BFCP, several alternatives were considered and explored to some
degree. Each of these is discussed briefly in the following
subsections. In summary, while these alternatives work in a number
of scenarios, they are not sufficient, in and of themselves, to
address the use case targeted by this draft.
3.1.1. ICE TCP
ICE TCP [I-D.ietf-mmusic-ice-tcp] extends ICE to TCP based media,
including the ability to offer a mix of TCP and UDP based candidates
for a single stream. ICE TCP has, in general, a lower success
probability for enabling TCP connectivity without a relay if both of
the hosts are behind a NAT (see Appendix A of
[I-D.ietf-mmusic-ice-tcp]) than enabling UDP connectivity in the same
scenarios. The happens because many of the currently deployed NATs
in video conferencing networks do not support the flow of TCP hand
shake packets seen in case of TCP simultaneous-open, either because
they do not allow incoming TCP SYN packets from an address to which a
SYN packet has been sent to recently, or because they do not properly
process the subsequent SYNACK. Implementing various techniques
advocated for candidate collection in [I-D.ietf-mmusic-ice-tcp]
should increase the success probability, but many of these techniques
require support from some network elements (e.g., from the NATs).
Such support is not common in enterprise firewalls and NATs.
3.1.2. Teredo
Teredo [RFC4380] enables nodes located behind one or more IPv4 NATs
to obtain IPv6 connectivity by tunneling packets over UDP. Teredo
extensions [RFC6081] provide additional capabilities to Teredo,
including support for more types of NATs and support for more
efficient communication.
As defined, Teredo could be used to make BFCP work for the video
conferencing use cases addressed in this draft. However, running the
service requires the help of "Teredo servers" and "Teredo relays"
[RFC4380]. These servers and relays generally do not exist in the
existing video conferencing deployments. It also requires IPv6
awareness on the endpoints. It should also be noted that ICMP6, as
used with Teredo to complete an initial protocol exchange and confirm
that the appropriate NAT bindings have been set up, is not a
conventional feature of IPv4 or even IPv6, and some currently
Kristensen, et al. Expires August 20, 2012 [Page 6]
Internet-Draft BFCP-Unreliable February 2012
deployed IPv6 firewalls discard ICMP messages. As these networks
continue to evolve and tackle the transaction to IPv6, Teredo servers
and relays may be deployed, making Teredo available as a suitable
alternative to BFCP over UDP.
3.1.3. GUT
GUT [I-D.manner-tsvwg-gut] attempts to facilitate tunneling over UDP
by encapsulating the native transport protocol and its payload (in
general the whole IP payload) within a UDP packet destined to the
well-known port GUT_P. Unfortunately, it requires user-space TCP, for
which there is not a readily available implementation, and creating
one is a large project in itself. This draft has expired and its
future is still not clear as it has not yet been adopted by a working
group.
3.1.4. UPnP IGD
Universal Plug and Play Internet Gateway Devices (UPnP IGD) sit on
the edge of the network, providing connectivity to the Internet for
computers internal to the LAN, but do not allow Internet devices to
connect to computers on the internal LAN. IGDs enable a computer on
an internal LAN to create port mappings on their NAT, through which
hosts on the Internet can send data that will be forwarded to the
computer on the internal LAN. IGDs may be self-contained hardware
devices or may be software components provided within an operating
system.
In considering UPnP IGD, several issues exist. Not all NATs support
UPnP, and many that do support it are configured with it turned off
by default. NATs are often multilayered, and UPnP does not work well
with such NATs. For example, a typical DSL modems acts as a NAT, and
the user plugs in a wireless access point behind that, which adds
another layer NAT. The client can discover the first layer of NAT
using multicast but it is harder to figure out how to discover and
control NATs in the next layer up.
3.1.5. NAT PMP
The NAT Port Mapping Protocol (NAT PMP) allows a computer in a
private network (behind a NAT router) to automatically configure the
router to allow parties outside the private network to contact it.
NAT PMP runs over UDP. It essentially automates the process of port
forwarding. Included in the protocol is a method for retrieving the
public IP address of a NAT gateway, thus allowing a client to make
this public IP address and port number known to peers that may wish
to communicate with it.
Kristensen, et al. Expires August 20, 2012 [Page 7]
Internet-Draft BFCP-Unreliable February 2012
Many NATs do not support PMP. In those that do support it, it has
similar issues with negotiation of multilayer NATs as UPnP. Video
conferencing is used extensively in enterprise networks, and NAT PMP
is not generally available in enterprise-class routers.
4. Difference from RFC4582
This section details the difference from [RFC4582], the base protocol
specification of BFCP, required for use over an unreliable transport.
The section numbers to which differences apply are indicated in
parentheses in the titles of the sub-sections below.
4.1. Overview of Operation (4)
Fourth paragraph change:
There are two types of transaction in BFCP: client-initiated
transactions and server-initiated transactions. Client-initiated
transactions consist of a message from a client to the floor
control server and a response from the floor control server to the
client. Correspondingly, server-initiated transactions consist of
a message from the floor control server to a client and the
associated acknowledgement message from the client to the floor
control server. Both messages can be related because they carry
the same Transaction ID value in their common headers.
4.1.1. Floor Participant to Floor Control Server Interface (4.1)
Before seventh paragraph (page 9), insert:
Figures 2 and 3 below show call flows for two sample BFCP
interactions when used over reliable transport. Appendix A
(Editorial Note: here-in Section 4.31) shows the same sample
interactions but over an unreliable transport.
4.2. COMMON-HEADER Format (5.1)
The figure below should replace Figure 5: COMMON-HEADER format.
Kristensen, et al. Expires August 20, 2012 [Page 8]
Internet-Draft BFCP-Unreliable February 2012
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Ver |I|F| Res | Primitive | Payload Length |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Conference ID |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Transaction ID | User ID |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Fragment Offset | Fragment Length |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Figure 2: COMMON-HEADER format
The description for "Ver" is changed as follows on page 15:
Ver: The 3-bit version field MUST be set to 1 when using BFCP over
reliable transport, i.e. as in [RFC4582]. The 3-bit version field
MUST be set to 2 when using BFCP over unreliable transport, with
the extensions specified in this document.
The following text precedes "Reserved" on page 15:
I: The Transaction Initiator (I) flag-bit has relevance only for
use of BFCP over unreliable transport. When cleared, it indicates
that this message is a request initiating a new transaction, and
the Transaction ID that follows has been generated for this
transaction. When set, it indicates that this message is a
response to a previous request, and the Transaction ID that
follows is the one associated with that request. When BFCP is
used over reliable transports, the flag has no significance and
SHOULD be cleared.
F: The Fragmentation (F) flag-bit has relevance only for use of
BFCP over unreliable transport. When cleared, the message is not
fragmented. When set, it indicates that the message is a fragment
of a large fragmented BFCP message. (The optional fields Fragment
Offset and Fragment Length described below are present only if the
F flag is set).
The Reserved field changes name to Res due to limited space in the
ASCII graphic in Figure 2. In the description of the Reserved field
"the 5 bits" is changed to "the 3 bits".
The description of Transaction ID should have the final clause
deleted with the reference to Section 8 remaining. The value used
for server-initiated transactions MUST be non-zero when BFCP is used
over unreliable transports, and this qualification shall be described
Kristensen, et al. Expires August 20, 2012 [Page 9]
Internet-Draft BFCP-Unreliable February 2012
in the updated Section 8.
The values below should be appended to the end of Table 1: BFCP
primitives.
+-------+-----------------------+--------------------+
| Value | Primitive | Direction |
+-------+-----------------------+--------------------+
| 14 | FloorRequestStatusAck | P -> S ; Ch -> S |
| 15 | ErrorAck | P -> S ; Ch -> S |
| 16 | FloorStatusAck | P -> S ; Ch -> S |
| 17 | Goodbye | P -> S ; Ch -> S ; |
| | | P <- S ; Ch <- S |
| 18 | GoodbyeAck | P -> S ; Ch -> S ; |
| | | P <- S ; Ch <- S |
+-------+-----------------------+--------------------+
Table 1: BFCP primitives
The following text should be added after "User ID" on page 15:
Fragment Offset: This optional field is present only if the F flag
is set and contains a 16-bit value that specifies the number of
4-octet units contained in previous fragments.
Fragment Length: This optional field is present only if the F flag
is set and contains a 16-bit value that specifies the number of
4-octet units contained in this fragment.
4.3. ERROR-CODE (5.2.6)
The value below should be appended to the end of Table 5: Error Code
meaning.
+-------+-------------------------+
| Value | Meaning |
+-------+-------------------------+
| 10 | Unable to parse message |
| 11 | Use DTLS |
+-------+-------------------------+
Table 2: Error Code meaning
4.4. FloorRequestStatusAck (5.3.14)
This new subsection specifies the normative ABNF for the new
primitive, FloorRequestStatusAck.
Kristensen, et al. Expires August 20, 2012 [Page 10]
Internet-Draft BFCP-Unreliable February 2012
Floor participants and chairs acknowledge the receipt of a
FloorRequestStatus message from the floor control server when
communicating over unreliable transport. The following is the
format of the FloorRequestStatusAck message:
FloorRequestStatusAck = (COMMON-HEADER)
*[EXTENSION-ATTRIBUTE]
Figure 3: FloorRequestStatusAck format
4.5. ErrorAck (5.3.15)
This new subsection specifies the normative ABNF for the new
primitive, ErrorAck.
Floor participants and chairs acknowledge the receipt of an Error
message from the floor control server when communicating over
unreliable transport. The following is the format of the ErrorAck
message:
ErrorAck = (COMMON-HEADER)
*[EXTENSION-ATTRIBUTE]
Figure 4: ErrorAck format
4.6. FloorStatusAck (5.3.16)
This new subsection specifies the normative ABNF for the new
primitive, FloorStatusAck.
Floor participants and chairs acknowledge the receipt of a
FloorStatus message from the floor control server when
communicating over unreliable transport. The following is the
format of the FloorStatusAck message:
FloorStatusAck = (COMMON-HEADER)
*[EXTENSION-ATTRIBUTE]
Figure 5: FloorStatusAck format
Kristensen, et al. Expires August 20, 2012 [Page 11]
Internet-Draft BFCP-Unreliable February 2012
4.7. Goodbye (5.3.17)
This new subsection specifies the normative ABNF for the new
primitive, Goodbye.
BFCP entities that wish to dissociate themselves from their remote
participant do so through the transmission of a Goodbye. The
following is the format of the Goodbye message:
Goodbye = (COMMON-HEADER)
*[EXTENSION-ATTRIBUTE]
Figure 6: Goodbye format
4.8. GoodbyeAck (5.3.18)
This new subsection specifies the normative ABNF for the new
primitive, GoodbyeAck.
BFCP entities communicating over an unreliable transport should
acknowledge the receipt of a Goodbye message from a peer. The
following is the format of the GoodbyeAck message:
GoodbyeAck = (COMMON-HEADER)
*[EXTENSION-ATTRIBUTE]
Figure 7: GoodbyeAck format
4.9. Transport (6)
An additional behavior is recommended for entities participating in
communication over an unreliable transport that either wish to leave
or are asked to leave an established BFCP connection, as detailed in
the revised section introduction text below.
The transport over which BFCP entities exchange messages depends
on how clients obtain information to contact the floor control
server (e.g. using an SDP offer/answer exchange [RFC4583]). Two
transports are supported: TCP, appropriate where entities can be
sure that their connectivity is not impeded by NAT devices, media
relays or firewalls; and UDP for those deployments where TCP may
not be applicable or appropriate.
Kristensen, et al. Expires August 20, 2012 [Page 12]
Internet-Draft BFCP-Unreliable February 2012
If a client wishes to end its BFCP association with a floor
control server, it is RECOMMENDED that the client send a Goodbye
message to dissociate itself from any allocated resources. If a
floor control server wishes to end its BFCP association with a
client (e.g. the Focus of the conference informs the floor control
server that the client has been kicked out from the conference),
it is RECOMMENDED that the floor control server send a Goodbye
message towards the client.
4.9.1. Reliable Transport (6.1)
BFCP entities may elect to exchange BFCP messages using TCP
connections. TCP provides an in-order reliable delivery of a stream
of bytes. Consequently, message framing is implemented in the
application layer. BFCP implements application-layer framing using
TLV-encoded attributes.
A client MUST NOT use more than one TCP connection to communicate
with a given floor control server within a conference. Nevertheless,
if the same physical box handles different clients (e.g. a floor
chair and a floor participant), which are identified by different
User IDs, a separate connection per client is allowed.
If a BFCP entity (a client or a floor control server) receives data
that cannot be parsed, the entity MUST close the TCP connection, and
the connection SHOULD be reestablished. Similarly, if a TCP
connection cannot deliver a BFCP message and times out, the TCP
connection SHOULD be reestablished.
The way connection reestablishment is handled depends on how the
client obtains information to contact the floor control server. Once
the TCP connection is reestablished, the client MAY resend those
messages for which it did not get a response from the floor control
server.
If a floor control server detects that the TCP connection towards one
of the floor participants is lost, it is up to the local policy of
the floor control server what to do with the pending floor requests
of the floor participant. In any case, it is RECOMMENDED that the
floor control server keep the floor requests (i.e., that it does not
cancel them) while the TCP connection is reestablished.
To maintain backwards compatibility with older implementations of
[RFC4583], BFCP entities MUST interpret the graceful close of their
TCP connection from their associated participant as an implicit
Goodbye message.
Kristensen, et al. Expires August 20, 2012 [Page 13]
Internet-Draft BFCP-Unreliable February 2012
4.9.2. Unreliable Transport (6.2)
BFCP entities may elect to exchange BFCP messages using UDP
datagrams. UDP is an unreliable transport where neither delivery nor
ordering is assured. Each BFCP UDP datagram MUST contain exactly one
BFCP message. In the event the size of a BFCP message exceeds the
MTU size, the BFCP message will be fragmented at the IP layer.
Considerations related to fragmentation are covered in Section 4.9.3.
The message format for exchange of BFCP in UDP datagrams is the same
as for a TCP stream above.
Clients MUST announce their presence to the floor control server by
transmission of a Hello message. This Hello message MUST be
responded to with a HelloAck message and only upon receipt can the
client consider the floor control service as present and available.
As described in Section 8, each request sent by a floor participant
or chair shall form a client transaction that expects an
acknowledgement message back from the floor control server within a
retransmission window. Concordantly, messages sent by the floor
control server that are not transaction-completing (e.g. FloorStatus
announcements as part of a FloorQuery subscription) are server-
initiated transactions that require acknowledgement messages from the
floor participant and chair entities to which they were sent.
If a BFCP entity receives data that cannot be parsed, the receiving
participant MAY send an Error message with parameter value 10
indicating receipt of a malformed message. If the message can be
parsed to the extent that it is able to discern that it was a
response to an outstanding request transaction, the client MAY
discard the message and await retransmission. BFCP entities
receiving an Error message with value 10 SHOULD acknowledge the error
and act accordingly.
Transaction ID values are non-sequential and entities are at liberty
to select values at random. Entities MUST only have at most one
outstanding request transaction at any one time. Implicit
subscriptions, such as FloorRequest messages that have multiple
responses as the floor control server processes intermediate states
until Granted or Denied terminal states attained, can be
characterized by a client-initiated request transaction whose
acknowledgement is implied by the first FloorRequestStatus response
from the floor control server. The subsequent changes in state for
the request are new transactions whose Transaction ID is determined
by the floor control server and whose receipt by the client
participant shall be acknowledged with a FloorRequestStatusAck
message. [Editorial note: would it be more straightforward to have
all FloorRequestStatus messages acknowledged with a
Kristensen, et al. Expires August 20, 2012 [Page 14]
Internet-Draft BFCP-Unreliable February 2012
FloorRequestStatusAck message?]
By restricting entities to having at most one pending transaction
open, both the out-of-order receipt of messages as well as the
possibility for congestion are mitigated. Additional details
regarding congestion control are provided in Section 4.9.2.1. A
server-initiated request (e.g. a FloorStatus with an update from the
floor control server) received by a participant before the initial
FloorRequestStatus message that closes the client-initiated
transaction that was instigated by the FloorRequest MUST be treated
as superseding the information conveyed in any delinquent response.
As the floor control server cannot send a second update to the
implicit floor status subscription until the first is acknowledged,
ordinality is maintained.
4.9.2.1. Congestion Control
BFCP may be characterized to generate "low data-volume" traffic, per
the classification in [RFC5405]. Nevertheless is it necessary to
ensure suitable and necessary congestion control mechanisms are used
for BFCP over UDP. As described in previous paragraph every entity -
client or server - is only allowed to send one request at a time, and
await the acknowledging response. This way at most one datagram is
sent per RTT given the message is not lost during transmission. In
case the message is lost, the request retransmission timer T1
specified in Section 4.14 will fire and the message is retransmitted
up to three times. The default initial interval is set to 500ms and
the interval is doubled after each retransmission attempt, this is
identical to the specification of the T1 timer in SIP as described in
Section 17.1.1.2 of [RFC3261].
4.9.2.2. ICMP Error Handling
If a BFCP entity receives an ICMP port unreachable message mid-
conversation, the entity SHOULD treat the conversation as closed
(e.g. an implicit Goodbye message from the peer) and behave
accordingly. The entity MAY attempt to re-establish the conversation
afresh. The new connection will appear as a wholly new floor
participant, chair or floor control server with all state previously
held about that participant lost.
Note: This is because the peer entities cannot rely on IP and port
tuple to uniquely identify the participant, nor would extending Hello
to include an attribute that advertised what the entity previously
was assigned as a User ID be acceptable due to session hijacking.
In deployments where NAT appliances, firewalls or other such devices
are present and affecting port reachability for each entity, one
Kristensen, et al. Expires August 20, 2012 [Page 15]
Internet-Draft BFCP-Unreliable February 2012
possibility is to utilize the peer connectivity checks, relay use and
NAT pinhole maintenance mechanisms defined in ICE [RFC5245].
4.9.3. Large Message Considerations
Large messages become a concern when using BFCP if the overall size
of a single BFCP message exceeds that representable within the 16-bit
Payload Length field of the COMMON-HEADER. When using UDP, there is
the added concern that a single BFCP message can be fragmented at the
IP layer if its overall size exceeds the MTU threshold of the
network.
4.9.3.1. Fragmentation Handling
When transmitting a BFCP message with size greater than the MTU, the
sender should fragment the message into a series of N contiguous data
ranges. The sender should then create N BFCP fragment messages (one
for each data range) with the same Transaction ID. The size of each
of these N messages MUST be smaller than the MTU. The F flag in the
COMMON-HEADER is set to indicate fragmentation of the BFCP message.
For each of these fragments the Fragment Offset and Fragment Length
fields are included in the COMMON-HEADER. The Fragment Offset field
denotes the number of bytes contained in the previous fragments. The
Fragment Length contains the length of the fragment itself. Note
that the Payload Length field contains the length of the entire,
unfragmented message.
When a BFCP implementation receives a BFCP message fragment, it MUST
buffer the fragment until it has received the entire BFCP message.
The state machine should handle the BFCP message only after all the
fragments for the message have been received.
If a fragment of a BFCP message is lost, the sender will not receive
an ACK for the message. Therefore the sender will retransmit the
message with same transaction ID as specified in Section 4.13. If
the ACK sent by the receiver is lost, then the entire message will be
resent by the sender. The receiver MUST then retransmit the ACK.
The receiver can discard an incomplete buffer utilizing the Response
Retransmission Timer, starting the timer after the receipt of the
first fragment.
4.10. Lower-Layer Security (7)
Expand the section to mandate support for DTLS when transport over
UDP is used such that it reads as follows:
Kristensen, et al. Expires August 20, 2012 [Page 16]
Internet-Draft BFCP-Unreliable February 2012
BFCP relies on lower-layer security mechanisms to provide replay
and integrity protection and confidentiality. BFCP floor control
servers and clients (which include both floor participants and
floor chairs) MUST support TLS for transport over TCP and MUST
support DTLS for transport over UDP [RFC5246]. Any BFCP entity
MAY support other security mechanisms.
BFCP entities MUST support, at a minimum, the
TLS_RSA_WITH_AES_128_CBC_SHA ciphersuite [RFC5246].
Which party, the client or the floor control server, acts as the
TLS/DTLS server depends on how the underlying TLS/DTLS connection
is established. For a TCP/TLS connection established using an SDP
offer/answer exchange [RFC4583], the answerer (which may be the
client or the floor control server) always acts as the TLS server.
For a UDP/DTLS connection established using the same exchange,
either party can be the DTLS server depending on the setup
attributes exchanged, as defined in [RFC5763].
4.11. Protocol Transactions (8)
The final clause of the introduction to section 8 should be read as:
Since they do not trigger any response, their Transaction ID is
set to 0 when used over reliable transports, but must be non-zero
and unique in the context of outstanding transactions over
unreliable transports.
When using BFCP over unreliable transports, all requests will use
retransmit timer T1 (see Section 4.13) until the transaction is
completed.
4.12. Server Behavior (8.2)
The final clause of this section should be read as:
Server-initiated transactions MUST contain a Transaction ID equal
to 0 when BFCP is used over reliable transports. Over unreliable
transport, the Transaction ID shall have the same properties as
for client-initiated transactions: the server MUST set the
Transaction ID value in the common header to a number that is
different from 0 and that MUST NOT be reused in another message
from the server until the appropriate response from the client is
received for the transaction. The server uses the Transaction ID
value to match this message with the response from the floor
participant or floor chair.
Kristensen, et al. Expires August 20, 2012 [Page 17]
Internet-Draft BFCP-Unreliable February 2012
4.13. Timers (8.3)
New section:
When BFCP entities are communicating over an unreliable transport,
two retransmission timers are employed to help mitigate against
loss of datagrams. Retransmission and response caching are not
required when BFCP entities communicate over reliable transports.
4.14. Request Retransmission Timer, T1 (8.3.1)
T1 is a timer that schedules retransmission of a request until an
appropriate response is received or until the maximum number of
retransmissions have occurred. The timer doubles on each re-
transmit, failing after three unacknowledged transmission attempts.
If a valid response is not received for a client- or server-initiated
transaction, the implementation MUST consider the BFCP association as
failed. Implementations SHOULD follow the reestablishment procedure
described in section 6 (e.g. initiate a new offer/answer [RFC3264]
exchange). Alternatively, they MAY continue without BFCP and
therefore not be participant in any floor control actions.
4.15. Response Retransmission Timer, T2 (8.3.2)
T2 is a timer that, when fires, signals that the BFCP entity can
release knowledge of the transaction against which it is running. It
is started upon the first transmission of the response to a request
and is the only mechanism by which that response is released by the
BFCP entity. Any subsequent retransmissions of the same request can
be responded to by replaying the cached response, whilst that value
is retained until the timer has fired.
T2 shall be set such that it encompasses all legal retransmissions
per T1 plus a factor to accommodate network latency between BFCP
entities.
4.16. Timer Values (8.3.3)
The table below defines the different timers required when BFCP
entities communicate over an unreliable transport.
Kristensen, et al. Expires August 20, 2012 [Page 18]
Internet-Draft BFCP-Unreliable February 2012
+-------+--------------------------------------+---------+
| Timer | Description | Value/s |
+-------+--------------------------------------+---------+
| T1 | Initial request retransmission timer | 0.5s |
| T2 | Response retransmission timer | 10s |
+-------+--------------------------------------+---------+
Table 3: Timers
The default value for T1 is 500 ms, this is an estimate of the RTT
for completing the transaction. T1 MAY be chosen larger, and this is
RECOMMENDED if it is known in advance that the RTT is larger.
Regardless of the value of T1, the exponential backoffs on
retransmissions described in Section 4.14 MUST be used.
4.17. Authentication and Authorization (9)
The first sentence of the second paragraph should be read as:
BFCP supports TLS/DTLS mutual authentication between client and
floor control servers, as specified in section 9.1.
4.17.1. TLS Based Mutual Authentication (9.1)
Change each instance of "TLS" to "TLS/DTLS", and each instance of
"TCP" to "TCP/UDP".
4.18. Receiving a Response [to a FloorRequest Message] (10.1.2)
Prepend the sentence below at the start of this subsection:
When communicating over unreliable transport and upon receiving a
FloorRequest from a participant, the floor control server MUST
respond with a FloorRequestStatus message within the transaction
failure window to complete the transaction.
4.19. Receiving a Response [to a FloorRelease Message] (10.2.2)
Prepend the sentence below at the start of this subsection:
When communicating over unreliable transport and upon receiving a
FloorRelease from a participant, the floor control server MUST
respond with a FloorRequestStatus message within the transaction
failure window to complete the transaction.
Kristensen, et al. Expires August 20, 2012 [Page 19]
Internet-Draft BFCP-Unreliable February 2012
4.20. Receiving a Response [to a ChairAction Message] (11.2)
Prepend the sentence below at the start of this subsection:
When communicating over unreliable transport and upon receiving a
ChairAction from a participant, the floor control server MUST
respond with a ChairActionAck message within the transaction
failure window to complete the transaction.
4.21. Receiving a Response [to a FloorQuery Message] (12.1.2)
Prepend the sentence below at the start of this subsection:
When communicating over unreliable transport and upon receiving a
FloorQuery from a participant, the floor control server MUST
respond with a FloorStatus message within the transaction failure
window to complete the transaction.
4.22. Receiving a Response [to a FloorRequestQuery Message] (12.2.2)
Prepend the sentence below at the start of this subsection:
When communicating over unreliable transport and upon receiving a
FloorRequestQuery from a participant, the floor control server
MUST respond with a FloorRequestStatus message within the
transaction failure window to complete the transaction.
4.23. Receiving a Response [to a UserQuery Message] (12.3.2)
Prepend the sentence below at the start of this subsection:
When communicating over unreliable transport and upon receiving a
UserQuery from a participant, the floor control server MUST
respond with a UserStatus message within the transaction failure
window to complete the transaction.
4.24. Receiving a Response [to a Hello Message] (12.4.2)
Prepend the sentence below at the start of this subsection:
When communicating over unreliable transport and upon receiving a
Hello from a participant, the floor control server MUST respond
with a HelloAck message within the transaction failure window to
complete the transaction.
Kristensen, et al. Expires August 20, 2012 [Page 20]
Internet-Draft BFCP-Unreliable February 2012
4.25. Reception of a FloorRequestStatus Message (13.1.3)
The sentence below shall appear as a new subsection:
When communicating over unreliable transport and upon receiving a
FloorRequestStatus message from a floor control server, the
participant MUST respond with a FloorRequestStatusAck message
within the transaction failure window to complete the transaction.
4.26. Reception of a FloorStatus Message (13.5.3)
The sentence below shall appear as a new subsection:
When communicating over unreliable transport and upon receiving a
FloorStatus message from a floor control server, the participant
MUST respond with a FloorStatusAck message within the transaction
failure window to complete the transaction.
4.27. Reception of an Error Message (13.8.1)
The sentence below shall appear as a new subsection:
When communicating over unreliable transport and upon receiving an
Error message from a floor control server, the participant MUST
respond with a ErrorAck message within the transaction failure
window to complete the transaction.
4.28. Security Considerations (14)
Change each instance of "TLS" to "TLS/DTLS", and each instance of
"TCP" to "TCP/UDP".
4.29. IANA Considerations - Primitive Subregistry (15.2)
This section instructs the IANA to register the following new values
for the BFCP primitive subregistry.
+-------+-----------------------+-------------+
| Value | Primitive | Reference |
+-------+-----------------------+-------------+
| 14 | FloorRequestStatusAck | RFC 4582bis |
| 15 | ErrorAck | RFC 4582bis |
| 16 | FloorStatusAck | RFC 4582bis |
| 17 | Goodbye | RFC 4582bis |
| 18 | GoodbyeAck | RFC 4582bis |
+-------+-----------------------+-------------+
Table 4: BFCP primitive subregistry
Kristensen, et al. Expires August 20, 2012 [Page 21]
Internet-Draft BFCP-Unreliable February 2012
4.30. IANA Considerations - Error Code Subregistry (15.4)
This section instructs the IANA to register the following new values
for the BFCP Error Code subregistry.
+-------+-------------------------+-------------+
| Value | Meaning | Reference |
+-------+-------------------------+-------------+
| 10 | Unable to parse message | RFC 4582bis |
| 11 | Use DTLS | RFC 4582bis |
+-------+-------------------------+-------------+
Table 5: BFCP Error Code subregistry
4.31. Example Call Flows for BFCP over Unreliable Transport (Appendix
A)
With reference to Section 4.1, the following figures show
representative call-flows for requesting and releasing a floor, and
obtaining status information about a floor when BFCP is deployed over
an unreliable transport. The figures here show a loss-less
interaction.
Editorial Note: A future version of this draft will show an example
with lost packets due to unreliable transport, as well as examples on
usage of DTLS and STUN in call the setup phase.
Floor Participant Floor Control
Server
|(1) FloorRequest |
|Transaction ID: 123 |
|User ID: 234 |
|FLOOR-ID: 543 |
|---------------------------------------------->|
| |
|(2) FloorRequestStatus |
|Transaction ID: 123 |
|User ID: 234 |
|FLOOR-REQUEST-INFORMATION |
| Floor Request ID: 789 |
| OVERALL-REQUEST-STATUS |
| Request Status: Pending |
| FLOOR-REQUEST-STATUS |
| Floor ID: 543 |
|<----------------------------------------------|
| |
|(3) FloorRequestStatus |
Kristensen, et al. Expires August 20, 2012 [Page 22]
Internet-Draft BFCP-Unreliable February 2012
|Transaction ID: 4098 |
|User ID: 234 |
|FLOOR-REQUEST-INFORMATION |
| Floor Request ID: 789 |
| OVERALL-REQUEST-STATUS |
| Request Status: Accepted |
| Queue Position: 1st |
| FLOOR-REQUEST-STATUS |
| Floor ID: 543 |
|<----------------------------------------------|
| |
|(4) FloorRequestStatusAck |
|Transaction ID: 4098 |
|User ID: 234 |
|---------------------------------------------->|
| |
|(5) FloorRequestStatus |
|Transaction ID: 4130 |
|User ID: 234 |
|FLOOR-REQUEST-INFORMATION |
| Floor Request ID: 789 |
| OVERALL-REQUEST-STATUS |
| Request Status: Granted |
| FLOOR-REQUEST-STATUS |
| Floor ID: 543 |
|<----------------------------------------------|
| |
|(6) FloorRequestStatusAck |
|Transaction ID: 4130 |
|User ID: 234 |
|---------------------------------------------->|
| |
|(7) FloorRelease |
|Transaction ID: 154 |
|User ID: 234 |
|FLOOR-REQUEST-ID: 789 |
|---------------------------------------------->|
| |
|(8) FloorRequestStatus |
|Transaction ID: 154 |
|User ID: 234 |
|FLOOR-REQUEST-INFORMATION |
| Floor Request ID: 789 |
| OVERALL-REQUEST-STATUS |
| Request Status: Released |
| FLOOR-REQUEST-STATUS |
| Floor ID: 543 |
|<----------------------------------------------|
Kristensen, et al. Expires August 20, 2012 [Page 23]
Internet-Draft BFCP-Unreliable February 2012
Figure 8: Requesting and releasing a floor
Note that in Figure 8, the FloorRequestStatus message from the floor
control server to the floor participant is a transaction-closing
message as a response to the client-initiated transaction with
Transaction ID 154. It does not and SHOULD NOT be followed by a
FloorRequestStatusAck message from the floor participant to the floor
control server.
Floor Participant Floor Control
Server
|(1) FloorQuery |
|Transaction ID: 257 |
|User ID: 234 |
|FLOOR-ID: 543 |
|---------------------------------------------->|
| |
|(2) FloorStatus |
|Transaction ID: 257 |
|User ID: 234 |
|FLOOR-ID:543 |
|FLOOR-REQUEST-INFORMATION |
| Floor Request ID: 764 |
| OVERALL-REQUEST-STATUS |
| Request Status: Accepted |
| Queue Position: 1st |
| FLOOR-REQUEST-STATUS |
| Floor ID: 543 |
| BENEFICIARY-INFORMATION |
| Beneficiary ID: 124 |
|FLOOR-REQUEST-INFORMATION |
| Floor Request ID: 635 |
| OVERALL-REQUEST-STATUS |
| Request Status: Accepted |
| Queue Position: 2nd |
| FLOOR-REQUEST-STATUS |
| Floor ID: 543 |
| BENEFICIARY-INFORMATION |
| Beneficiary ID: 154 |
|<----------------------------------------------|
| |
|(3) FloorStatus |
|Transaction ID: 4319 |
|User ID: 234 |
|FLOOR-ID:543 |
|FLOOR-REQUEST-INFORMATION |
| Floor Request ID: 764 |
Kristensen, et al. Expires August 20, 2012 [Page 24]
Internet-Draft BFCP-Unreliable February 2012
| OVERALL-REQUEST-STATUS |
| Request Status: Granted |
| FLOOR-REQUEST-STATUS |
| Floor ID: 543 |
| BENEFICIARY-INFORMATION |
| Beneficiary ID: 124 |
|FLOOR-REQUEST-INFORMATION |
| Floor Request ID: 635 |
| OVERALL-REQUEST-STATUS |
| Request Status: Accepted |
| Queue Position: 1st |
| FLOOR-REQUEST-STATUS |
| Floor ID: 543 |
| BENEFICIARY-INFORMATION |
| Beneficiary ID: 154 |
|<----------------------------------------------|
| |
|(4) FloorStatusAck |
|Transaction ID: 4319 |
|User ID: 234 |
|---------------------------------------------->|
| |
|(5) FloorStatus |
|Transaction ID: 4392 |
|User ID: 234 |
|FLOOR-ID:543 |
|FLOOR-REQUEST-INFORMATION |
| Floor Request ID: 635 |
| OVERALL-REQUEST-STATUS |
| Request Status: Granted |
| FLOOR-REQUEST-STATUS |
| Floor ID: 543 |
| BENEFICIARY-INFORMATION |
| Beneficiary ID: 154 |
|<----------------------------------------------|
| |
|(6) FloorStatusAck |
|Transaction ID: 4392 |
|User ID: 234 |
|---------------------------------------------->|
Figure 9: Obtaining status information about a floor
5. Revision of RFC4583
This section details revisions to [RFC4583], the SDP format for
specifying BFCP streams. The section number to which updates apply
Kristensen, et al. Expires August 20, 2012 [Page 25]
Internet-Draft BFCP-Unreliable February 2012
are indicated in parentheses in the titles of the sub-sections below.
5.1. Fields in the 'm' Line (3)
The section shall be re-written to remove reference to the
exclusivity of TCP as a transport for BFCP streams.
1. In paragraph four, "... will initiate its TCP connection ..."
becomes "... will direct BFCP messages ..."
2. In paragraph four, delete "Since BFCP only runs on top of TCP,
the port is always a TCP port."
3. Change paragraph five, "We define two new values ... ", to, "We
define four new values for the transport field: TCP/BFCP, TCP/
TLS/BFCP, UDP/BFCP, and UDP/TLS/BFCP. TCP/BFCP is used when BFCP
runs directly on top of TCP, and TCP/TLS/BFCP is used when BFCP
runs on top of TLS, which in turn runs on top of TCP. Similarly,
UDP/BFCP is used when BFCP runs directly on top of UDP, and UDP/
TLS/BFCP is used when BFCP runs on top of DTLS [RFC4347], which
in turn runs on top of UDP."
5.2. Authentication (8)
In last paragraph, change "When TLS is used, once the underlaying TCP
connection is established" to "When TLS is used with TCP, once the
underlying connection is established".
5.3. Security Considerations (10)
Append to the first paragraph, "Furthermore, when using DTLS over
UDP, considerations for its use with RTP and RTCP are presented in
[RFC5763]. The requirements for the offer/answer exchange, as listed
in Section 5 of that document, MUST be followed."
5.4. Registration of SDP 'proto' Values (11.1)
This section should be renamed now that there are more values to
register in the SDP parameters registry, with the following added to
the table:
+--------------+-------------+
| Value | Reference |
+--------------+-------------+
| UDP/BFCP | RFC 4583bis |
| UDP/TLS/BFCP | RFC 4583bis |
+--------------+-------------+
Kristensen, et al. Expires August 20, 2012 [Page 26]
Internet-Draft BFCP-Unreliable February 2012
Table 6: Value for the SDP 'proto' field
6. NAT Traversal
One of the key benefits when using UDP for BFCP communication is the
ability to leverage the existing NAT traversal infrastructure and
strategies deployed to facilitate transport of the media associated
with the video conferencing sessions. Depending on the given
deployment, this infrastructure typically includes some subset of ICE
[RFC5245].
In order to facilitate the initial establishment of NAT bindings, and
to maintain those bindings once established, BFCP over UDP entities
are RECOMMENDED to use STUN [RFC5389] for keep-alives, as described
for SIP [RFC5626]. This results in each BFCP entity sending a
packet, both to open the pinhole and to learn what IP/port the NAT
assigned for the binding.
In order to facilitate traversal of BFCP packets through NATs, BFCP
over UDP entities are RECOMMENDED to use symmetric ports for sending
and receiving BFCP packets, as recommended for RTP/RTCP [RFC4961].
7. Remaining Work
This draft reflects a work in progress, with at least the following
items to be documented and/or revised:
Example signaling flows: A later version of this draft will include
further examples - as appropriate - of signaling exchanges over
unreliable transport as a visual aid and reference for
implementers, potential candidates: Updated transactions,
message retransmission, usage of DTLS during call setup, and
combined usage of DTLS and STUN.
Reformat and merge: After figuring out the technical details in this
draft, the "diff" will be merged to form proper bis-drafts to
become RFC4582bis (in BFCPbis WG) and RFC4583bis (in MMUSIC
WG).
Other issues not related to transport Fixing erratas to the RFCs and
known minor issues with the existing specification.
Kristensen, et al. Expires August 20, 2012 [Page 27]
Internet-Draft BFCP-Unreliable February 2012
8. Contributing Authors
The authors/editors would like to thank Mark K. Thompson, Eoin McLeod
and Nivedita Melinkeri who made a major contribution to the
development of this document.
Eoin McLeod
Cisco
Email: eoimcleo@cisco.com
Nivedita Melinkeri
Cisco
Email: nivedita@cisco.com
Mark K. Thompson
Cisco
Email: markth2@cisco.com
9. Acknowledgements
We acknowledge contributions to one or more previous versions of this
draft from Trond G. Andersen, Gonzalo Camarillo, Roni Even, Lorenzo
Miniero, Joerg Ott, Hadriel Kaplan, Dan Wing, Cullen Jennings, David
Benham, and Alan Ford.
10. References
10.1. Normative References
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119, March 1997.
[RFC3264] Rosenberg, J. and H. Schulzrinne, "An Offer/Answer Model
with Session Description Protocol (SDP)", RFC 3264,
June 2002.
[RFC4347] Rescorla, E. and N. Modadugu, "Datagram Transport Layer
Security", RFC 4347, April 2006.
[RFC4582] Camarillo, G., Ott, J., and K. Drage, "The Binary Floor
Control Protocol (BFCP)", RFC 4582, November 2006.
[RFC4583] Camarillo, G., "Session Description Protocol (SDP) Format
for Binary Floor Control Protocol (BFCP) Streams",
RFC 4583, November 2006.
Kristensen, et al. Expires August 20, 2012 [Page 28]
Internet-Draft BFCP-Unreliable February 2012
[RFC4961] Wing, D., "Symmetric RTP / RTP Control Protocol (RTCP)",
BCP 131, RFC 4961, July 2007.
[RFC5246] Dierks, T. and E. Rescorla, "The Transport Layer Security
(TLS) Protocol Version 1.2", RFC 5246, August 2008.
[RFC5389] Rosenberg, J., Mahy, R., Matthews, P., and D. Wing,
"Session Traversal Utilities for NAT (STUN)", RFC 5389,
October 2008.
[RFC5626] Jennings, C., Mahy, R., and F. Audet, "Managing Client-
Initiated Connections in the Session Initiation Protocol
(SIP)", RFC 5626, October 2009.
10.2. Informative References
[I-D.ietf-mmusic-ice-tcp]
Rosenberg, J., Keranen, A., Lowekamp, B., and A. Roach,
"TCP Candidates with Interactive Connectivity
Establishment (ICE)", draft-ietf-mmusic-ice-tcp-16 (work
in progress), November 2011.
[I-D.ietf-mmusic-media-path-middleboxes]
Stucker, B. and H. Tschofenig, "Analysis of Middlebox
Interactions for Signaling Protocol Communication along
the Media Path",
draft-ietf-mmusic-media-path-middleboxes-03 (work in
progress), July 2010.
[I-D.manner-tsvwg-gut]
Manner, J., Varis, N., and B. Briscoe, "Generic UDP
Tunnelling (GUT)", draft-manner-tsvwg-gut-02 (work in
progress), July 2010.
[IMC05] Guha, S. and P. Francis, "Characterization and Measurement
of TCP Traversal through NATs and Firewalls", 2005,
.
[P2PNAT] Ford, B., Srisuresh, P., and D. Kegel, "Peer-to-Peer
Communication Across Network Address Translators",
April 2005,
.
[RFC3261] Rosenberg, J., Schulzrinne, H., Camarillo, G., Johnston,
A., Peterson, J., Sparks, R., Handley, M., and E.
Schooler, "SIP: Session Initiation Protocol", RFC 3261,
June 2002.
Kristensen, et al. Expires August 20, 2012 [Page 29]
Internet-Draft BFCP-Unreliable February 2012
[RFC4380] Huitema, C., "Teredo: Tunneling IPv6 over UDP through
Network Address Translations (NATs)", RFC 4380,
February 2006.
[RFC5245] Rosenberg, J., "Interactive Connectivity Establishment
(ICE): A Protocol for Network Address Translator (NAT)
Traversal for Offer/Answer Protocols", RFC 5245,
April 2010.
[RFC5405] Eggert, L. and G. Fairhurst, "Unicast UDP Usage Guidelines
for Application Designers", BCP 145, RFC 5405,
November 2008.
[RFC5763] Fischl, J., Tschofenig, H., and E. Rescorla, "Framework
for Establishing a Secure Real-time Transport Protocol
(SRTP) Security Context Using Datagram Transport Layer
Security (DTLS)", RFC 5763, May 2010.
[RFC6081] Thaler, D., "Teredo Extensions", RFC 6081, January 2011.
Appendix A. Change History
A.1. draft-ietf-bfcpbis-rfc4582bis-00 to -01
1. Mandated using version 2 (Ver field == 2) for BFCP over UDP, with
the extensions described in this draft. For BFCP over TCP, the
version is still 1.
2. Added text regarding fragmentation handling: A new 'F' flag and
Fragment Offset field in Section 4.2. Added fragmentation
handling mechanism in Section 4.9.3.1.
3. Resolve an inconsistency between Section 4.10 and Section 5.3, by
introducing the setup attribute for DTLS.
4. Moved some authors to the new Section 8 Contributing Authors.
5. A dash of editorial polish.
A.2. draft-sandbakken-dispatch-bfcp-udp-03 to
draft-ietf-bfcpbis-rfc4582bis-00
1. Draft name change. Adopted as main work item in BFCPbis WG.
2. Switched from informational to standards track.
Kristensen, et al. Expires August 20, 2012 [Page 30]
Internet-Draft BFCP-Unreliable February 2012
3. No conflict with IANA registries for BFCP, since the aim is a
standards track RFC. Removed text in Future work section.
4. Just editorial changes as requested by WG chairs; used as a
starting point in the new WG. Will add changes in upcoming
version. Also author list will be considered, for instance
adding a contributors section in the draft.
A.3. draft-sandbakken-dispatch-bfcp-udp-02 to -03
1. Added fragmentation and reassembly mechanism defined for RELOAD
as a candidate mechanism for consideration for BFCP when
transported over UDP.
2. Added ERROR-CODE to indicate DTLS is required.
3. Added UDP/TLS/BFCP as 4th transport value for BFCP.
4. Added requirement to follow offer/answer procedure in [RFC5763]
when using DTLS over UDP for BFCP.
A.4. draft-sandbakken-dispatch-bfcp-udp-01 to -02
1. Switched from standards track to informational.
2. Added section on motivation, including alternatives considered,
to address issues raised at IETF 79 and on various workgroup
aliases.
3. Changed semantics of the Transaction Initiator (I) flag-bit.
4. Expanded transport section to more explicitly call out
considerations regarding congestion control and ICMP errors, and
add considerations for large messages.
5. Updated security related sections and added authentication
section to address DTLS when using UDP.
6. Added section on NAT Traversal.
7. Some editorial changes.
A.5. draft-sandbakken-dispatch-bfcp-udp-00 to -01
1. Decision made to not increase the protocol version number as a
result of this extension. Certain aspects of this draft require
different behaviors depending on whether a reliable or unreliable
transport is being used, e.g. server-initiated transactions
Kristensen, et al. Expires August 20, 2012 [Page 31]
Internet-Draft BFCP-Unreliable February 2012
having Transaction ID 0 over reliable transports without
acknowledgements versus non-zero and active-unique with an
acknowledgement message when entities communicate over unreliable
transports. As the graceful-close behavior of [RFC4582] is still
allowed for TCP-based implementations without mandating the use
of the new Goodbye message, there is no need to change the
version number.
2. Removed the - a bit too verbose - rationale/motivation text
describing background and why other approaches where not chosen.
Was OK for a -00 draft, not strictly needed.
3. Not mandate ICE as a SHALL, but leave it as a non-mandatory way
of solving the potential need for NAT/FW traversal.
4. Emphasized that the reference to DTLS-SRTP are merely
informational.
5. A dash of polish and nitpicking added, some typos fixed.
A.6. draft-sandbakken-xcon-bfcp-udp-02 to
draft-sandbakken-dispatch-bfcp-udp-00
1. Draft name change. As XCON WG is closing this draft is submitted
to Dispatch WG as the arena of discussion.
2. Moved Transaction Identifier bit (I) from the Transaction ID to
one of the current 5 reserved bits. Keep current Transaction ID
syntax and semantics. Avoid potential problems with existing TCP
based implementations.
3. The way congestion control is taken care of is explained, with
reference to [RFC5405]. One message per RTT. Backoff and
normative behavior for timer T1 clarified.
4. Mandated support for DTLS in case unreliable transport (i.e.
UDP) is implemented. Details and examples to be included. Model
after [RFC5763], details on how to adapt the SRTP associated
details to BFCP and whether a reference or copying the text
across and changing is needed.
5. Added the Rationale and Scope section to position and explain the
motivation for this draft more in detail.
6. A number of typos and editorial changes.
Kristensen, et al. Expires August 20, 2012 [Page 32]
Internet-Draft BFCP-Unreliable February 2012
A.7. draft-sandbakken-xcon-bfcp-udp-01 to -02
1. Stepped away from changing semantics and directionality of Hello
and HelloAck messages for pinhole establishment and keep-alive in
favor of ICE toolset, particularly as this would have not
resolved connectivity establishment as a precursor to deployment
of DTLS [RFC4347] as a transport security mechanism.
2. Change to COMMON-HEADER to reserve bit-16 of Transaction ID to
show originator of transaction such that request/response and
response/acknowledgement mapping can be maintained without
colliding randomly chosen Transaction IDs. This also avoids a
three-way handshake scenario around FloorRequest where the
implicit acknowledgement (in FloorRequestStatus) might also be
interpreted as a transaction opening request on the part of the
floor control server.
3. Defined additional timer (T2) to soak up lost responses without
additional processing.
4. Restricted outstanding transactions to only one in-flight per
direction at any one time to mitigate re-ordering issues.
5. Defined entity behavior when transactions timeout.
6. Specified initial suggestion for how to minimize fragmentation of
messages.
7. Removed consideration of TCP-over-UDP after internal review.
8. Re-stated DTLS as likely preferred mechanism of securing
transport, although this investigation is on-going.
A.8. draft-sandbakken-xcon-bfcp-udp-00 to -01
1. Refactored to a format that represents explicit changes to base
RFCs.
2. Introduction of issues currently under investigation that
preclude adoption.
3. Specified retransmission timer for requests.
Kristensen, et al. Expires August 20, 2012 [Page 33]
Internet-Draft BFCP-Unreliable February 2012
Authors' Addresses
Tom Kristensen (editor)
Cisco
Philip Pedersens vei 22
N-1366 Lysaker
Norway
Email: tomkrist@cisco.com, tomkri@ifi.uio.no
Charles Eckel
Cisco
170 West Tasman Drive
San Jose, CA 95134
United States
Email: eckelcu@cisco.com
Alfred E. Heggestad
Cisco
Philip Pedersens vei 22
N-1366 Lysaker
Norway
Email: aheggest@cisco.com
Geir A. Sandbakken
Cisco
Philip Pedersens vei 22
N-1366 Lysaker
Norway
Email: geirsand@cisco.com
Kristensen, et al. Expires August 20, 2012 [Page 34]