Network Working Group                                     A. Gulbrandsen
Request for Comments: 4978                        Oryx Mail Systems GmbH
Category: Standards Track                                    August 2007


                      The IMAP COMPRESS Extension

Status of this Memo

   This document specifies an Internet standards track protocol for the
   Internet community, and requests discussion and suggestions for
   improvements.  Please refer to the current edition of the "Internet
   Official Protocol Standards" (STD 1) for the standardization state
   and status of this protocol.  Distribution of this memo is unlimited.

Abstract

   The COMPRESS extension allows an IMAP connection to be effectively
   and efficiently compressed.

Table of Contents

   1. Introduction and Overview .......................................2
   2. Conventions Used in This Document ...............................2
   3. The COMPRESS Command ............................................3
   4. Compression Efficiency ..........................................4
   5. Formal Syntax ...................................................6
   6. Security Considerations .........................................6
   7. IANA Considerations .............................................6
   8. Acknowledgements ................................................7
   9. References ......................................................7
      9.1. Normative References .......................................7
      9.2. Informative References .....................................7

1.  Introduction and Overview

   A server which supports the COMPRESS extension indicates this with
   one or more capability names consisting of "COMPRESS=" followed by a
   supported compression algorithm name as described in this document.

   The goal of COMPRESS is to reduce the bandwidth usage of IMAP.

   Compared to PPP compression (see [RFC1962]) and modem-based
   compression (see [MNP] and [V42BIS]), COMPRESS offers much better
   compression efficiency.  COMPRESS can be used together with Transport
   Security Layer (TLS) [RFC4346], Simple Authentication and Security
   layer (SASL) encryption, Virtual Private Networks (VPNs), etc.
   Compared to TLS compression [RFC3749], COMPRESS has the following
   (dis)advantages:

   - COMPRESS can be implemented easily both by IMAP servers and
     clients.

   - IMAP COMPRESS benefits from an intimate knowledge of the IMAP
     protocol's state machine, allowing for dynamic and aggressive
     optimization of the underlying compression algorithm's parameters.

   - When the TLS layer implements compression, any protocol using that
     layer can transparently benefit from that compression (e.g., SMTP
     and IMAP).  COMPRESS is specific to IMAP.

   In order to increase interoperation, it is desirable to have as few
   different compression algorithms as possible, so this document
   specifies only one.  The DEFLATE algorithm (defined in [RFC1951]) is
   standard, widely available and fairly efficient, so it is the only
   algorithm defined by this document.

   In order to increase interoperation, IMAP servers that advertise this
   extension SHOULD also advertise the TLS DEFLATE compression mechanism
   as defined in [RFC3749].  IMAP clients MAY use either COMPRESS or TLS
   compression, however, if the client and server support both, it is
   RECOMMENDED that the client choose TLS compression.

   The extension adds one new command (COMPRESS) and no new responses.

2.  Conventions Used in This Document

   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].

   Formal syntax is defined by [RFC4234] as modified by [RFC3501].

   In the examples, "C:" and "S:" indicate lines sent by the client and
   server respectively. "[...]" denotes elision.

3.  The COMPRESS Command

   Arguments: Name of compression mechanism: "DEFLATE".

   Responses: None

   Result: OK The server will compress its responses and expects the
              client to compress its commands.
           NO Compression is already active via another layer.
          BAD Command unknown, invalid or unknown argument, or COMPRESS
              already active.

   The COMPRESS command instructs the server to use the named
   compression mechanism ("DEFLATE" is the only one defined) for all
   commands and/or responses after COMPRESS.

   The client MUST NOT send any further commands until it has seen the
   result of COMPRESS.  If the response was OK, the client MUST compress
   starting with the first command after COMPRESS.  If the server
   response was BAD or NO, the client MUST NOT turn on compression.

   If the server responds NO because it knows that the same mechanism is
   active already (e.g., because TLS has negotiated the same mechanism),
   it MUST send COMPRESSIONACTIVE as resp-text-code (see [RFC3501],
   Section 7.1), and the resp-text SHOULD say which layer compresses.

   If the server issues an OK response, the server MUST compress
   starting immediately after the CRLF which ends the tagged OK
   response.  (Responses issued by the server before the OK response
   will, of course, still be uncompressed.)  If the server issues a BAD
   or NO response, the server MUST NOT turn on compression.

   For DEFLATE (as for many other compression mechanisms), the
   compressor can trade speed against quality.  When decompressing there
   isn't much of a tradeoff.  Consequently, the client and server are
   both free to pick the best reasonable rate of compression for the
   data they send.

   When COMPRESS is combined with TLS (see [RFC4346]) or SASL (see
   [RFC4422]) security layers, the sending order of the three extensions
   MUST be first COMPRESS, then SASL, and finally TLS.  That is, before
   data is transmitted it is first compressed.  Second, if a SASL
   security layer has been negotiated, the compressed data is then
   signed and/or encrypted accordingly.  Third, if a TLS security layer
   has been negotiated, the data from the previous step is signed and/or

   encrypted accordingly.  When receiving data, the processing order
   MUST be reversed.  This ensures that before sending, data is
   compressed before it is encrypted, independent of the order in which
   the client issues COMPRESS, AUTHENTICATE, and STARTTLS.

   The following example illustrates how commands and responses are
   compressed during a simple login sequence:

        S: * OK [CAPABILITY IMAP4REV1 STARTTLS COMPRESS=DEFLATE]
        C: a starttls
        S: a OK TLS active

            From this point on, everything is encrypted.

        C: b login arnt tnra
        S: b OK Logged in as arnt
        C: c compress deflate
        S: d OK DEFLATE active

            From this point on, everything is compressed before being
            encrypted.

   The following example demonstrates how a server may refuse to
   compress twice:

        S: * OK [CAPABILITY IMAP4REV1 STARTTLS COMPRESS=DEFLATE]
        [...]
        C: c compress deflate
        S: c NO [COMPRESSIONACTIVE] DEFLATE active via TLS

4.  Compression Efficiency

   This section is informative, not normative.

   IMAP poses some unusual problems for a compression layer.

   Upstream is fairly simple.  Most IMAP clients send the same few
   commands again and again, so any compression algorithm that can
   exploit repetition works efficiently.  The APPEND command is an
   exception; clients that send many APPEND commands may want to
   surround large literals with flushes in the same way as is
   recommended for servers later in this section.

   Downstream has the unusual property that several kinds of data are
   sent, confusing all dictionary-based compression algorithms.

   One type is IMAP responses.  These are highly compressible; zlib
   using its least CPU-intensive setting compresses typical responses to
   25-40% of their original size.

   Another type is email headers.  These are equally compressible, and
   benefit from using the same dictionary as the IMAP responses.

   A third type is email body text.  Text is usually fairly short and
   includes much ASCII, so the same compression dictionary will do a
   good job here, too.  When multiple messages in the same thread are
   read at the same time, quoted lines etc. can often be compressed
   almost to zero.

   Finally, attachments (non-text email bodies) are transmitted, either
   in binary form or encoded with base-64.

   When attachments are retrieved in binary form, DEFLATE may be able to
   compress them, but the format of the attachment is usually not IMAP-
   like, so the dictionary built while compressing IMAP does not help.
   The compressor has to adapt its dictionary from IMAP to the
   attachment's format, and then back.  A few file formats aren't
   compressible at all using deflate, e.g., .gz, .zip, and .jpg files.

   When attachments are retrieved in base-64 form, the same problems
   apply, but the base-64 encoding adds another problem.  8-bit
   compression algorithms such as deflate work well on 8-bit file
   formats, however base-64 turns a file into something resembling 6-bit
   bytes, hiding most of the 8-bit file format from the compressor.

   When using the zlib library (see [RFC1951]), the functions
   deflateInit2(), deflate(), inflateInit2(), and inflate() suffice to
   implement this extension.  The windowBits value must be in the range
   -8 to -15, or else deflateInit2() uses the wrong format.
   deflateParams() can be used to improve compression rate and resource
   use.  The Z_FULL_FLUSH argument to deflate() can be used to clear the
   dictionary (the receiving peer does not need to do anything).

   A client can improve downstream compression by implementing BINARY
   (defined in [RFC3516]) and using FETCH BINARY instead of FETCH BODY.
   In the author's experience, the improvement ranges from 5% to 40%
   depending on the attachment being downloaded.

   A server can improve downstream compression if it hints to the
   compressor that the data type is about to change strongly, e.g., by
   sending a Z_FULL_FLUSH at the start and end of large non-text
   literals (before and after '*CHAR8' in the definition of literal in
   RFC 3501, page 86).  Small literals are best left alone.  A possible
   boundary is 5k.

   A server can improve the CPU efficiency both of the server and the
   client if it adjusts the compression level (e.g., using the
   deflateParams() function in zlib) at these points, to avoid trying to
   compress incompressible attachments.  A very simple strategy is to
   change the level to 0 at the start of a literal provided the first
   two bytes are either 0x1F 0x8B (as in deflate-compressed files) or
   0xFF 0xD8 (JPEG), and to keep it at 1-5 the rest of the time.  More
   complex strategies are possible.

5.  Formal Syntax

   The following syntax specification uses the Augmented Backus-Naur
   Form (ABNF) notation as specified in [RFC4234].  This syntax augments
   the grammar specified in [RFC3501].  [RFC4234] defines SP and
   [RFC3501] defines command-auth, capability, and resp-text-code.

   Except as noted otherwise, all alphabetic characters are case-
   insensitive.  The use of upper or lower case characters to define
   token strings is for editorial clarity only.  Implementations MUST
   accept these strings in a case-insensitive fashion.

       command-auth =/ compress

       compress    = "COMPRESS" SP algorithm

       capability  =/ "COMPRESS=" algorithm
                     ;; multiple COMPRESS capabilities allowed

       algorithm   = "DEFLATE"

       resp-text-code =/ "COMPRESSIONACTIVE"

   Note that due the syntax of capability names, future algorithm names
   must be atoms.

6.  Security Considerations

   As for TLS compression [RFC3749].

7.  IANA Considerations

   The IANA has added COMPRESS=DEFLATE to the list of IMAP capabilities.

8.  Acknowledgements

   Eric Burger, Dave Cridland, Tony Finch, Ned Freed, Philip Guenther,
   Randall Gellens, Tony Hansen, Cullen Jennings, Stephane Maes, Alexey
   Melnikov, Lyndon Nerenberg, and Zoltan Ordogh have all helped with
   this document.

   The author would also like to thank various people in the rooms at
   meetings, whose help is real, but not reflected in the author's
   mailbox.

9.  References

9.1.  Normative References

   [RFC1951]  Deutsch, P., "DEFLATE Compressed Data Format Specification
              version 1.3", RFC 1951, May 1996.

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119, March 1997.

   [RFC3501]  Crispin, M., "INTERNET MESSAGE ACCESS PROTOCOL - VERSION
              4rev1", RFC 3501, March 2003.

   [RFC4234]  Crocker, D. and P. Overell, "Augmented BNF for Syntax
              Specifications: ABNF", RFC 4234, October 2005.

9.2.  Informative References

   [RFC1962]  Rand, D., "The PPP Compression Control Protocol (CCP)",
              RFC 1962, June 1996.

   [RFC3516]  Nerenberg, L., "IMAP4 Binary Content Extension", RFC 3516,
              April 2003.

   [RFC3749]  Hollenbeck, S., "Transport Layer Security Protocol
              Compression Methods", RFC 3749, May 2004.

   [RFC4346]  Dierks, T. and E. Rescorla, "The Transport Layer Security
              (TLS) Protocol Version 1.1", RFC 4346, April 2006.

   [RFC4422]  Melnikov, A. and  K. Zeilenga, "Simple Authentication and
              Security Layer (SASL)", RFC 4422, June 2006.

   [V42BIS]   ITU, "V.42bis: Data compression procedures for data
              circuit-terminating equipment (DCE) using error correction
              procedures", http://www.itu.int/rec/T-REC-V.42bis, January
              1990.

   [MNP]      Gilbert Held, "The Complete Modem Reference", Second
              Edition, Wiley Professional Computing, ISBN 0-471-00852-4,
              May 1994.

Author's Address

    Arnt Gulbrandsen
    Oryx Mail Systems GmbH
    Schweppermannstr. 8
    D-81671 Muenchen
    Germany

    Fax: +49 89 4502 9758
    EMail: arnt@oryx.com

Full Copyright Statement

   Copyright (C) The IETF Trust (2007).

   This document is subject to the rights, licenses and restrictions
   contained in BCP 78, and except as set forth therein, the authors
   retain all their rights.

   This document and the information contained herein are provided on an
   "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS
   OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND
   THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS
   OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF
   THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED
   WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Intellectual Property

   The IETF takes no position regarding the validity or scope of any
   Intellectual Property Rights or other rights that might be claimed to
   pertain to the implementation or use of the technology described in
   this document or the extent to which any license under such rights
   might or might not be available; nor does it represent that it has
   made any independent effort to identify any such rights.  Information
   on the procedures with respect to rights in RFC documents can be
   found in BCP 78 and BCP 79.

   Copies of IPR disclosures made to the IETF Secretariat and any
   assurances of licenses to be made available, or the result of an
   attempt made to obtain a general license or permission for the use of
   such proprietary rights by implementers or users of this
   specification can be obtained from the IETF on-line IPR repository at
   http://www.ietf.org/ipr.

   The IETF invites any interested party to bring to its attention any
   copyrights, patents or patent applications, or other proprietary
   rights that may cover technology that may be required to implement
   this standard.  Please address the information to the IETF at
   ietf-ipr@ietf.org.