multi0.nr

.\" format this document with "rfc multi1.nr draft-aboba-avtcore-quic-multiplexing-01.txt"
.nr HY 0
.pl 10.0i
.po 0
.ll 7.2i
.lt 7.2i
.nr LL 7.2i
.nr LT 7.2i
.ds LH INTERNET-DRAFT
.ds CH QUIC Multiplexing
.ds RH \*(DY
.ds LF Aboba, et. al
.ds CF Informational
.ds RF FORMFEED[Page %]

.tl 'AVTCORE Working Group''B. Aboba'
.tl 'INTERNET-DRAFT''Microsoft Corporation'
.tl 'Category: Informational''P. Thatcher'
.tl 'Expires: April 29, 2018''Google'
.tl '''C. Perkins'
.tl '''University of Glasgow'
.tl '''\*(DY'
.ad l
.LP
.ce
QUIC Multiplexing
.ce
draft-aboba-avtcore-quic-multiplexing-01.txt

Abstract

.in +0.3i
If QUIC is to be used in a peer-to-peer manner, with NAT traversal,
then it is necessary to be able to demultiplex QUIC and STUN flows
running on a single UDP port.  This memo discusses options for how
to perform such demultiplexing.  It also considers demultiplexing
of QUIC and WebRTC traffic (both media and data) when running on a 
single UDP port.
.in -0.3i

Status of This Memo

.in +0.3i
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 April 29, 2018.
.in -0.3i
.bp
Copyright Notice

.in +0.3i
Copyright (c) 2017 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.
.in -0.3i

Table of Contents

.in +0.3i
.nf
1.  Introduction . . . . . . . . . . . . . . . . . . . . . . . . .  3
  1.1.  Terminology  . . . . . . . . . . . . . . . . . . . . . . .  4
2.  Solutions  . . . . . . . . . . . . . . . . . . . . . . . . . .  4
  2.1.  QUIC Header Changes  . . . . . . . . . . . . . . . . . . .  4
  2.2.  Multiplexing Shim  . . . . . . . . . . . . . . . . . . . .  5
  2.3.  Heuristics . . . . . . . . . . . . . . . . . . . . . . . .  5
3.  Security Considerations  . . . . . . . . . . . . . . . . . . .  6
4.  IANA Considerations  . . . . . . . . . . . . . . . . . . . . .  7
5.  References . . . . . . . . . . . . . . . . . . . . . . . . . .  7
  5.1. Informative references  . . . . . . . . . . . . . . . . . .  7
Acknowledgments  . . . . . . . . . . . . . . . . . . . . . . . . .  9
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . .  9
.fi
.in -0.3i
.bp
.NH 1
.R
Introduction

.in +0.3i
QUIC [I-D.ietf-quic-transport] is a new network transport protocol.
While it is initially intended as a replacement for TCP in order to
better support HTTP/2 [RFC7540] it should eventually be useful as a
general purpose transport. HTTP is an asymmetric client-server 
protocol, but other uses of QUIC might operate in a peer-to-peer manner
and so will need effective NAT traversal using ICE [RFC5245], which
which makes use of STUN [RFC5389] and TURN [RFC5766] to discover NAT bindings.
This STUN and TURN traffic needs to run on the same UDP port
as the QUIC traffic. Accordingly, if QUIC is to be used in a peer-to-peer
manner, then it needs to be possible to demultiplex QUIC, STUN, and TURN
traffic running on a single UDP port. This memo discusses how to do this.

In addition, there are a number of ways in which communication between
WebRTC peers may utilize QUIC. One of these is transport of
RTP over QUIC, described in [I-D.rtpfolks-quic-rtp-over-quic].
Another is use of QUIC for data exchange. A Javascript API for
use of QUIC in WebRTC data exchange has been incorporated into the 
ORTC API [ORTC], under development within the W3C ORTC Community Group. 

In a WebRTC scenario where ICE is utilized for NAT traversal, 
SRTP [RFC3711] is keyed using DTLS-SRTP [RFC5764] and QUIC is used for data exchange, 
RTP/RTCP [RFC3550], STUN, TURN, DTLS [RFC6347], ZRTP [RFC6189] and QUIC
may all need to be multiplexed over a single ICE transport.

As noted in [RFC7983] Figure 3, protocol demultiplexing currently
relies upon differentiation based on the first octet, as follows:

.nf
                    +----------------+
                    |        [0..3] -+--> forward to STUN
                    |                |
                    |      [16..19] -+--> forward to ZRTP
                    |                |
        packet -->  |      [20..63] -+--> forward to DTLS
                    |                |
                    |      [64..79] -+--> forward to TURN Channel
                    |                |
                    |    [128..191] -+--> forward to RTP/RTCP
                    +----------------+

     Figure 1: DTLS-SRTP receiver's packet demultiplexing algorithm.
.fi

As noted by Colin Perkins and Lars Eggert in [QUIC-Issue] this creates a potential conflict
with the current design of the QUIC headers described in [I-D.ietf-quic-transport], since
the first octet of the QUIC header is either:

.nf
 +-+-+-+-+-+-+-+-+
 |1|   Type (7)  |  Long header packet
 +-+-+-+-+-+-+-+-+
.fi

which potentially produces values of the first octet in the range 129-134, conflicting
with RTP/RTCP, or

.nf
 +-+-+-+-+-+-+-+-+
 |0|C|K| Type (5)|  Short header packet
 +-+-+-+-+-+-+-+-+
.fi

which produces values for the first octet in the ranges 1-3, 33-35, 65-67 or 97-99,
potentially conflicting with STUN, DTLS and TURN.
.in -0.3i
.NH 2
.R
Terminology

.in +0.3i
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].
.in -0.3i
.NH 1
.R
Solutions

.in +0.3i
This section presents potential solutions to the QUIC multiplexing problem,
including changes to the QUIC headers, addition of a multiplexing octet and
use of heuristics.
.in -0.3i
.NH 2
.R
QUIC Header Changes

.in +0.3i
As noted in [QUIC-Issue], one potential solution involves changes to the QUIC headers,
such as setting the top two bits of the first octet of a QUIC packet to 1.  This
would imply a reduction in the size of the type fields: 

.nf
 +-+-+-+-+-+-+-+-+
 |1|1|1|Type (5) |  Long header packet
 +-+-+-+-+-+-+-+-+

 +-+-+-+-+-+-+-+-+
 |1|1|0|C|K|Type3|  Short header packet
 +-+-+-+-+-+-+-+-+
.fi

Note: [QUIC-Spin] proposes to add a spin bit to the type octet within the QUIC header, in order to allow for RTT calculation.
This would leave 4 bits for the type field in the long header packet and 2 bits for the type field in the short header, which
would accommodate the type field values allocated in [I-D.ietf-quic-transport]. 

.in +0.3i
The advantage to this approach is that it adds no additional overhead on-the-wire.
However it does require a reduction in the size of the QUIC Type fields and could potentially
require allocation of the following initial octet code points for QUIC: For the Long header,
225-230 (241-246 when the spin bit is set) and for the Short header,
193-195 (209-11 with spin bit set), 209-211 (225-227 with spin bit set) and 217-219 (233-235 with the spin bit set).
Utilizing all of these code points for QUIC would leave limited code points available for future allocations.
.in -0.3i
.NH 2
.R
Multiplexing Shim

.in +0.3i
In this approach, an initial octet not allocated within [RFC7983] would be prepended to
each QUIC packet, allowing QUIC packets to be differentiated from RTP, RTCP, DTLS, STUN,
TURN and ZRTP based on the first octet alone. As an example, an octet with decimal
value 192 could be used:

.nf
 +-+-+-+-+-+-+-+-+
 |1|1|0|0|0|0|0|0|
 +-+-+-+-+-+-+-+-+
.fi

.in +0.3i
Advantages of this approach include simplicity and the consumption of
only a single initial octet code point for demultiplexing
of QUIC. The disadvantage is the addition of a single octet of overhead to every QUIC packet, which
could impact performance where small payloads are exchanged, such as in peer-to-peer gaming. 
.in -0.3i
.NH 2
.R
Heuristics

.in +0.3i
During the QUIC WG interim in Seattle, Martin Thomson suggested the following heuristics for differentiation
of QUIC packets from RTP/RTCP/DTLS/STUN/TURN/ZRTP: 

.nf
1. Demultiplex differently during the "QUIC handshake"
   and "steady state".  
2. During handshake, we only need to worry about the QUIC
   Long header, which simplifies the logic.
    a. Force all handshake packets to utilize the QUIC Long header.
    b. The QUIC Long header (0x1XXXXXXX) (or 0x11XXXXXX with
       the spin bit set) does not conflict with STUN (0x000000XX),
       DTLS (0x000XXXXX), or TURN Channel (0x01XXXXXX).
    c. The QUIC Long header does conflict with RTP/RTCP (0x10XXXXXX),
       but those packets typically aren't sent until the QUIC
       handshake is completed. Corner case: an application starts
       off with audio and video keyed with DTLS-SRTP without QUIC,
       then the application wishes to add QUIC data (e.g. the user
       clicks on the "white-board" icon).  
        i. Alternative: force the RTP padding bit to 1
           using a one-byte pad if there isn't already
           padding (pad == 0x01).  Then force QUIC to have
           a type < 64 (the current max is 8).
       ii. Alternative: Disallow QUIC in this case, use SCTP data
           exchange instead.
3. During "steady state", we only need to worry about the QUIC 
   Short header. 
    a. QUIC doesn't need the Long header after the handshake.
    b. The QUIC Short header (0x0XXXXXXX or 0x01XXXXXX with
       the spin bit set) does not conflict with RTP/RTCP
       (0x10XXXXXX), so we only need to worry about
       conflicts with STUN/TURN/DTLS/ZRTP.
    c. Disallow simultaneous use of DTLS and QUIC
       Short header packets.
         i. Alternative: when using DTLS and QUIC at the same 
            time, only use the QUIC Long header. Not optimal,
            but isn't really needed.
    d. ICE can be demultiplexed using the magic cookie
       and checksum.
         i. Alternative: STUN can only conflict with 3
            QUIC packet types: Version Negotiation,
            Client Initial, and Server Stateless Retry.
            Out of those, none should be needed during
            the steady state. 
    e. We shouldn't need to demultiplex QUIC with TURN channel
       data or other STUN traffic. What about consent packets?
.fi

.in +0.3i
This approach has the advantage that it requires no changes to QUIC headers,
nor does it add any overhead to QUIC packets. Disadvantages include additional
complexity within the multiplexing algorithm, the consumption
of additional multiplexing code points, and potential future difficulties in
adapting the algorithm to support changes to the QUIC protocol or
additional protocols to be multiplexed.
.in -0.3i
.NH 1
.R
Security Considerations

.in +0.3i
The solutions discussed in this document could potentially introduce some
additional security considerations beyond those detailed in [RFC7983].

Due to the additional logic required, if mis-implemented,
heuristics have the potential to mis-classify packets.

When QUIC is used for only for data exchange, the TLS-within-QUIC exchange [I-D.ietf-quic-tls]
derives keys used solely to protect the QUIC data
packets.  If properly implemented, this should not affect the transport of SRTP nor the derivation of SRTP keys via DTLS-SRTP,
but if badly implemented, both transport and key derivation could be adversely impacted. 
.in -0.3i
.NH 1
.R
IANA Considerations

.in +0.3i
This document does not require actions by IANA.
.in -0.3i
.NH 1
.R
References
.NH 2
.R
Informative References
.IP [I-D.ietf-quic-tls] 13
Thomson, M. and S. Turner, "Using Transport Layer Security
(TLS) to Secure QUIC", draft-ietf-quic-tls-07
(work in progress), October 13, 2017.
.IP [I-D.ietf-quic-transport]
Iyengar, J. and M. Thomson, "QUIC: A UDP-Based Multiplexed
and Secure Transport", draft-ietf-quic-transport-07
(work in progress), October 13, 2017.
.IP [I-D.rtpfolks-quic-rtp-over-quic]
Ott, J., Even, R., Perkins, C. and V. Singh,
"RTP over QUIC", draft-rtpfolks-quic-rtp-over-quic-01
(work in progress), September 1, 2017.
.IP [ORTC]
Raymond, R., Aboba, B. and J. Uberti,
"Object RTC (ORTC) API for WebRTC", W3C,
http://draft.ortc.org/, October 2017.
.IP [QUIC-Issue]
Perkins, C., "QUIC header format/demultiplexing",
https://github.com/quicwg/base-drafts/issues/426,
March, 2017.
.IP [QUIC-Spin]
Huitema, C., "QUIC Latency Spin Bit",
https://github.com/quicwg/base-drafts/issues/609, June 2017.
.IP [RFC2119]
Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119,
DOI 10.17487/RFC2119, March 1997,
<http://www.rfc-editor.org/info/rfc2119>.
.IP [RFC3550]
Schulzrinne, H., Casner, S., Frederick, R., and V.
Jacobson, "RTP: A Transport Protocol for Real-Time
Applications", STD 64, RFC 3550, DOI 10.17487/RFC3550,
July 2003, <http://www.rfc-editor.org/info/rfc3550>.
.IP [RFC3711]
Baugher, M., McGrew, D., Naslund, M., Carrara, E., and K.
Norrman, "The Secure Real-time Transport Protocol (SRTP)",
RFC 3711, DOI 10.17487/RFC3711, March 2004,
<http://www.rfc-editor.org/info/rfc3711>.
.IP [RFC5245]
Rosenberg, J., "Interactive Connectivity Establishment
(ICE): A Protocol for Network Address Translator (NAT)
Traversal for Offer/Answer Protocols", RFC 5245,
DOI 10.17487/RFC5245, April 2010,
<http://www.rfc-editor.org/info/rfc5245>.
.IP [RFC5389]
Rosenberg, J., Mahy, R., Matthews, P., and D. Wing,
"Session Traversal Utilities for NAT (STUN)", RFC 5389,
DOI 10.17487/RFC5389, October 2008,
<http://www.rfc-editor.org/info/rfc5389>.
.IP [RFC5764]  
McGrew, D. and E. Rescorla, "Datagram Transport Layer
Security (DTLS) Extension to Establish Keys for the Secure
Real-time Transport Protocol (SRTP)", RFC 5764,
DOI 10.17487/RFC5764, May 2010,
<http://www.rfc-editor.org/info/rfc5764>.
.IP [RFC5766]
Mahy, R., Matthews, P., and J. Rosenberg, "Traversal Using
Relays around NAT (TURN): Relay Extensions to Session
Traversal Utilities for NAT (STUN)", RFC 5766,
DOI 10.17487/RFC5766, April 2010,
<http://www.rfc-editor.org/info/rfc5766>.
.IP [RFC6189]
Zimmermann, P., Johnston, A., Ed., and J. Callas, "ZRTP:
Media Path Key Agreement for Unicast Secure RTP",
RFC 6189, DOI 10.17487/RFC6189, April 2011,
<http://www.rfc-editor.org/info/rfc6189>.
.IP [RFC6347]
Rescorla, E. and N. Modadugu, "Datagram Transport Layer
Security Version 1.2", RFC 6347, DOI 10.17487/RFC6347,
January 2012, <http://www.rfc-editor.org/info/rfc6347>.
.IP [RFC7540]
Belshe, M., Peon, R., and M. Thomson, Ed., "Hypertext
Transfer Protocol Version 2 (HTTP/2)", RFC 7540,
DOI 10.17487/RFC7540, May 2015,
<https://www.rfc-editor.org/info/rfc7540>.
.IP [RFC7983]
Petit-Huguenin, M. and G. Salgueiro, 
"Multiplexing Scheme Updates
for Secure Real-time Transport Protocol (SRTP) Extension
for Datagram Transport Layer Security (DTLS)",
RFC 7983, DOI 10.17487/RFC7983, September 2016,
<https://www.rfc-editor.org/info/rfc7983>.
.LP
.bp
Acknowledgments

.in +0.3i
We would like to thank Martin Thomson, Roni Even and other participants
in the IETF QUIC and AVTCORE working groups
for their discussion of the QUIC multiplexing
issue, and their input relating to potential solutions.
.in -0.3i

Authors' Addresses

.in +0.3i
.nf
Bernard Aboba
Microsoft Corporation
One Microsoft Way
Redmond, WA  98052
USA

Email:  bernard.aboba@gmail.com

Peter Thatcher
Google
747 6th St S
Kirkland, WA  98033
USA

Email: pthatcher@google.com

Colin Perkins
School of Computing Science
University of Glasgow
Glasgow  G12 8QQ
United Kingdom

Email: csp@csperkins.org
.fi
.in -0.3i