RFC 4236






Network Working Group                                        A. Rousskov
Request for Comments: 4236                       The Measurement Factory
Category: Standards Track                                     M. Stecher
                                                  CyberGuard Corporation
                                                           November 2005


        HTTP Adaptation with Open Pluggable Edge Services (OPES)

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.

Copyright Notice



   Copyright (C) The Internet Society (2005).

Abstract



   Open Pluggable Edge Services (OPES) framework documents several
   application-agnostic mechanisms such as OPES tracing, OPES bypass,
   and OPES callout protocol.  This document extends those generic
   mechanisms for Hypertext Transfer Protocol (HTTP) adaptation.
   Together, application-agnostic OPES documents and this HTTP profile
   constitute a complete specification for HTTP adaptation with OPES.






















Rousskov & Stecher          Standards Track                     [Page 1]

RFC 4236               HTTP Adaptation with OPES           November 2005


Table of Contents



   1. Scope ...........................................................3
   2. OPES Document Map ...............................................3
   3. Callout Protocol ................................................4
      3.1. Application Message Parts ..................................5
      3.2. Application Profile Features ...............................6
           3.2.1. Profile Parts .......................................6
           3.2.2. Profile Structure ...................................8
           3.2.3. Aux-Parts ...........................................8
           3.2.4. Pause-At-Body .......................................9
           3.2.5. Stop-Receiving-Body ................................10
           3.2.6. Preservation-Interest-Body .........................10
           3.2.7. Content-Encodings ..................................11
           3.2.8. Profile Negotiation Example ........................12
      3.3. Application Message Start Message .........................13
      3.4. DUM Message ...............................................13
      3.5. Selective Adaptation ......................................14
      3.6. Hop-by-hop Headers ........................................15
      3.7. Transfer Encodings ........................................15
      3.8. HTTP Header Correctness ...................................16
           3.8.1. Message Size Recalculation .........................16
           3.8.2. Content-MD5 Header .................................17
      3.9. Examples ..................................................18
   4. Tracing ........................................................22
   5. Bypass .........................................................24
   6. IAB Considerations .............................................24
   7. Security Considerations ........................................24
   8. IANA Considerations ............................................24
   9. Compliance .....................................................25
   10. References ....................................................25
      10.1. Normative References .....................................25
      10.2. Informative References ...................................25


















Rousskov & Stecher          Standards Track                     [Page 2]

RFC 4236               HTTP Adaptation with OPES           November 2005


1.  Scope



   The Open Pluggable Edge Services (OPES) framework documents several
   application-agnostic mechanisms such as OPES processor and endpoints
   communications [RFC3897] or OPES callout protocol [RFC4037].  This
   document extends those generic mechanisms for adaptation of a
   specific application protocol, HTTP [RFC2616].  Together,
   application-agnostic OPES documents and this HTTP profile constitute
   a complete specification for HTTP adaptation with OPES.

   The primary sections of this document specify HTTP-specific
   extensions for the corresponding application-agnostic mechanisms
   documented elsewhere.

2.  OPES Document Map



   This document belongs to a large set of OPES specifications produced
   by the IETF OPES Working Group.  Familiarity with the overall OPES
   approach and typical scenarios is often essential when trying to
   comprehend isolated OPES documents.  This section provides an index
   of OPES documents to assist the reader with finding "missing"
   information.

   o  The document on "OPES Use Cases and Deployment Scenarios"
      [RFC3752] describes a set of services and applications that are
      considered in scope for OPES and have been used as a motivation
      and guidance in designing the OPES architecture.

   o  The OPES architecture and common terminology are described in "An
      Architecture for Open Pluggable Edge Services (OPES)" [RFC3835].

   o  "Policy, Authorization and Enforcement Requirements of OPES"
      [RFC3838] outlines requirements and assumptions on the policy
      framework, without specifying concrete authorization and
      enforcement methods.

   o  "Security Threats and Risks for OPES" [RFC3837] provides OPES risk
      analysis, without recommending specific solutions.

   o  "OPES Treatment of IAB Considerations" [RFC3914] addresses all
      architecture-level considerations expressed by the IETF Internet
      Architecture Board (IAB) when the OPES WG was chartered.

   o  At the core of the OPES architecture are the OPES processor and
      the callout server, two network elements that communicate with
      each other via an OPES Callout Protocol (OCP).  The requirements
      for such protocol are discussed in "Requirements for OPES Callout
      Protocols" [RFC3836].



Rousskov & Stecher          Standards Track                     [Page 3]

RFC 4236               HTTP Adaptation with OPES           November 2005


   o  "OPES Callout Protocol Core" [RFC4037] specifies an application
      agnostic protocol core to be used for the communication between
      OPES processor and callout server.

   o  "OPES entities and end points communications" [RFC3897] specifies
      generic tracing and bypass mechanisms for OPES.

   o  The OCP Core and Communications documents are independent from the
      application protocol being adapted by OPES entities.  Their

      generic mechanisms have to be complemented by application-specific
      profiles.  This document, HTTP adaptation with OPES, is such an
      application profile for HTTP.  It specifies how application-
      agnostic OPES mechanisms are to be used and augmented in order to
      support adaptation of HTTP messages.

   o  Finally, "P: Message Processing Language" [rules-p] defines a
      language for specifying what OPES adaptations (e.g., translation)
      must be applied to what application messages (e.g., e-mail from
      bob@example.com).  P language is meant for configuring application
      proxies (OPES processors).

3.  Callout Protocol



   This section documents the HTTP profile for the OPES Callout Protocol
   (OCP) Core [RFC4037].  Familiarity with OCP Core is required to
   understand the HTTP profile.  This section uses OCP Core conventions,
   terminology, and mechanisms.

   OPES processor communicates its desire to adapt HTTP messages via a
   Negotiation Offer (NO) message with HTTP-specific feature identifiers
   documented in Section 3.2.  HTTP-specific OCP optimization mechanisms
   can be negotiated at the same time.  A callout server that supports
   adaptation of HTTP messages has a chance to negotiate what HTTP
   message parts will participate in adaptation, including negotiation
   of HTTP request parts as metadata for HTTP response adaptation.
   Negotiable HTTP message parts are documented in Section 3.1.

   HTTP profile introduces a new parameter for the Application Message
   Start (AMS) message to communicate known HTTP message length (HTTP
   headers often do not convey length information reliably or at all).
   This parameter is documented in Section 3.3.  Section 3.4 documents a
   mechanism to report HTTP message parts with Data Use Mine (DUM)
   messages.

   The remaining OCP sections document various OCP marshaling corner
   cases such as handling of HTTP transfer encodings and 100 Continue
   responses.



Rousskov & Stecher          Standards Track                     [Page 4]

RFC 4236               HTTP Adaptation with OPES           November 2005


3.1.  Application Message Parts



   An HTTP message may have several well-known parts: headers, body, and
   trailers.  HTTP OPES processors are likely to have information about
   HTTP message parts because they have to isolate and interpret HTTP
   headers and find HTTP message boundaries.  Callout servers may either
   not care about certain parts or may benefit from reusing HTTP OPES
   processor work on isolating and categorizing interesting parts.

   The following is the declaration of am-part (application message
   part) type using OCP Core Protocol Element Type Declaration Mnemonic
   (PETDM):

   am-part:  extends atom;
   am-parts: extends list of am-part;

                                 Figure 1

   The following six "am-part" atoms are valid values:

   request-header: The start-line of an HTTP request message, all
      request message headers, and the CRLF separator at the end of HTTP
      headers (compare with section 4.1 of [RFC2616]).

   request-body: The message body of an HTTP request message as defined
      in section 4.3 of [RFC2616] but not including the trailer.

   request-trailer: The entity headers of the trailer of an HTTP request
      message in chunked transfer encoding.  This part follows the same
      syntax as the trailer defined in section 3.6.1 of [RFC2616].

   response-header: The start-line of an HTTP response message, all
      response message headers, and the CRLF separator at the end of
      HTTP headers (compare with section 4.1 of [RFC2616]).

   response-body: The message body of an HTTP response message as
      defined in section 4.3 of [RFC2616] but not including the trailer.

   response-trailer: The entity headers of the trailer of an HTTP
      response message in chunked transfer encoding.  This part follows
      the same syntax as the trailer defined in section 3.6.1 of
      [RFC2616].









Rousskov & Stecher          Standards Track                     [Page 5]

RFC 4236               HTTP Adaptation with OPES           November 2005


3.2.  Application Profile Features



   This document defines two HTTP profiles for OCP: request and response
   profiles.  These two profiles are described below.  Each profile has
   a unique feature identifier, a list of original application message
   parts, and a list of adapted application message parts:

   profile ID: http://www.iana.org/assignments/opes/ocp/http/request

      original request parts: request-header, request-body, request-
         trailer

      adapted request parts: request-header, request-body, request-
         trailer

      adapted response parts: response-header, response-body, response-
         trailer

   profile ID: http://www.iana.org/assignments/opes/ocp/http/response

      original transaction parts: request-header (aux), request-body
         (aux), request-trailer (aux), response-header, response-body,
         response-trailer

      adapted response parts: response-header, response-body, response-
         trailer

   The request profile contains two variants of adapted part lists: HTTP
   request parts and HTTP response parts.  Parts marked with an "(aux)"
   suffix are auxiliary parts that can only be used if explicitly
   negotiated for a profile.  See Section 3.2.1 for specific rules
   governing negotiation and use of am-parts.

   The scope of a negotiated profile is the OCP connection (default) or
   the service group specified via the SG parameter.

3.2.1.  Profile Parts



   An OCP agent MUST send application message parts in the order implied
   by the profile parts lists above.  An OCP agent receiving an out-of-
   order part MAY terminate the transaction with an error.

   An OPES processor MUST NOT send parts that are not listed as
   "original" in the negotiated profile.  A callout server MUST NOT send
   parts that are not listed as "adapted" in the negotiated profile.  An
   OCP agent receiving an not-listed part MUST terminate the transaction
   with an error.  The informal rationale for the last requirement is to
   reduce the number of subtle interoperability problems where an agent



Rousskov & Stecher          Standards Track                     [Page 6]

RFC 4236               HTTP Adaptation with OPES           November 2005


   thinks that the parts it is sending are understood/used by the other
   agent when, in fact, they are being ignored or skipped because they
   are not expected.

   Some HTTP messages lack certain parts.  For example, many HTTP
   requests do not have bodies, and most HTTP messages do not have
   trailers.  An OCP agent MUST NOT send (i.e., must skip) absent
   application message parts.

   An OCP agent MUST send present non-auxiliary parts and it MUST send
   those present auxiliary parts that were negotiated via the Aux-Parts
   (Section 3.2.3) parameter.  OCP agents MUST NOT send auxiliary parts
   that were not negotiated via the Aux-Parts (Section 3.2.3) parameter.

   An OCP agent receiving a message part in violation of the above
   requirements MAY terminate the corresponding transaction with an
   error.

   By design, original parts not included in the adapted parts list
   cannot be adapted.  In other words, a callout service can only adapt
   parts in the adapted parts list even though it may have access to
   other parts.

   In the request profile, the callout server MUST send either adapted

   request parts or adapted response parts.  An OPES processor receiving
   adapted flow with application message parts from both lists (in
   violation of the previous rule) MUST terminate the OCP transaction
   with an error.  Informally, the callout server sends adapted response
   parts to "short-circuit" the HTTP transaction, forcing the OPES
   processor to return an HTTP response without forwarding an adapted
   HTTP request.  This short-circuiting is useful for responding, for
   example, to an HTTP request that the callout service defines as
   forbidden.

   Unless explicitly configured to do otherwise, an OPES processor MUST
   offer all non-auxiliary original parts in Negotiation Offer (NO)
   messages.  See Section 3.5 for this rule rationale and examples of
   harmful side-effects from selective adaptation.












Rousskov & Stecher          Standards Track                     [Page 7]

RFC 4236               HTTP Adaptation with OPES           November 2005


3.2.2.  Profile Structure



   An HTTP application profile feature extends semantics of the feature
   type of OCP Core while adding the following named parameters to that
   type:

   o  Aux-Parts (Section 3.2.3)

   o  Pause-At-Body (Section 3.2.4)

   o  Stop-Receiving-Body (Section 3.2.5)

   o  Preservation-Interest-Body (Section 3.2.6)

   o  Content-Encodings (Section 3.2.7)

   The definition of the HTTP profile feature structure using PETDM
   follows:

   HTTP-Profile: extends Feature with {
       [Aux-Parts: am-parts];
       [Pause-At-Body: size];
       [Stop-Receiving-Body: size];
       [Preservation-Interest-Body: size];
       [Content-Encodings: codings];
   };

                                 Figure 2

   An HTTP profile structure can be used in feature lists of Negotiation
   Offer (NO) messages and as an anonymous parameter of a Negotiation
   Response (NR) message.  All profile parameters apply to any OCP
   transaction within profile scope.

3.2.3.  Aux-Parts



   The Aux-Parts parameter of an HTTP response profile can be used to
   negotiate the inclusion of auxiliary application message parts into
   the original data flow.  The parameter is a possibly empty list of
   am-part tokens.  An OPES processor MAY send an Aux-Parts parameter to
   advertise availability of auxiliary application message parts.  A
   callout server MAY respond with a possibly empty subset of the parts
   it needs.  The callout server response defines the subset of
   successfully negotiated auxiliary message parts.

   When receiving a Negotiation Offer (NO) message, the callout server
   MUST ignore any non-auxiliary part listed in the Aux-Parts parameter.
   When sending a Negotiation Response (NR) message, the callout server



Rousskov & Stecher          Standards Track                     [Page 8]

RFC 4236               HTTP Adaptation with OPES           November 2005


   MUST NOT select any application message part that was not explicitly
   listed in the negotiation offer.  In case of a violation of the last
   rule, the OPES processor MUST terminate the transaction.

   An OPES processor MUST send each negotiated auxiliary part to the
   callout server, unless the part is absent.

   Example:
        Aux-Parts: (request-header,request-body)

                                 Figure 3

3.2.4.  Pause-At-Body



   A callout server MAY use the Pause-At-Body parameter to request a
   pause in original application message body transmission before
   original dataflow starts.  The parameter's value is of type "offset".
   The parameter specifies the start of the non-auxiliary application
   message body suffix that the sender is temporarily not interested in
   seeing.

   [headers][ body prefix | body suffix ][trailer]
   <-- ? --><-- offset  --><-- ? ---------------->
   <-- equiv. DWP offset ->

                                 Figure 4

   When an OPES processor receives a Pause-At-Body parameter, it MUST
   behave as if it has received a Want Data Paused (DWP) message with
   the corresponding org-offset.  Note that the latter offset is
   different from the Pause-At-Body offset and is unknown until the size
   of the HTTP message headers is known.

   For example, if the Pause-At-Body value is zero, the OPES processor
   should send a Paused My Data (DPM) message just before it sends the
   first Data Use Mine (DUM) message with the response-body part in the
   HTTP response profile.  If the Pause-At-Body value is 300, the OPES
   processor should send a DPM message after transmitting 300 OCTETs for
   that application message part.

   Example:
        Pause-At-Body: 0

                                 Figure 5







Rousskov & Stecher          Standards Track                     [Page 9]

RFC 4236               HTTP Adaptation with OPES           November 2005


3.2.5.  Stop-Receiving-Body



   A callout server MAY use the Stop-Receiving-Body parameter to imply a
   Want Stop Receiving Data (DWSR) message behavior before the original
   dataflow starts.  The parameter's value is of type "offset".  The
   parameter specifies an offset into the original, non-auxiliary
   message body part (request-body in request profile and response-body
   in response profile).

   A callout service MAY send a Stop-Receiving-Body parameter with its
   negotiation response if there is a fixed offset into the message body
   for all transactions of a profile for which a Want Stop Receiving
   Data (DWSR) message would be sent.  An OPES processor MUST behave as
   if it has received a DWSR message with the corresponding offset.
   Note that the latter offset is different from the Stop-Receiving-Body
   offset and is unknown until the size of the HTTP message headers is
   known.

   For example, if the Stop-Receiving-Body value is zero in an HTTP
   response profile, the OPES processor should send an Application
   Message End (AME) message with result code 206 immediately after
   sending the response-header message part and before starting with the
   response-body message part.

   Example:
       Stop-Receiving-Body: 0

                                 Figure 6

3.2.6.  Preservation-Interest-Body



   The Preservation-Interest-Body parameter can be used to optimize data
   preservation at the OPES processor.  The parameter's value is of type
   "size" and denominates a prefix size of the original, non-auxiliary
   message body part (request-body in HTTP request profile and
   response-body in response profile).

   A callout service MAY send a Preservation-Interest-Body parameter
   with its negotiation response if there is a fixed-size prefix of the
   application message body for which a Data Preservation Interest (DPI)
   message would be sent.  An OPES processor MUST behave as if it
   receives a DPI message with org-offset zero and org-size equal to the
   value of the Preservation-Interest-Body parameter.








Rousskov & Stecher          Standards Track                    [Page 10]

RFC 4236               HTTP Adaptation with OPES           November 2005


   For example, if the Preservation-Interest-Body value is zero in an
   HTTP response profile, the callout server must not send any Data Use
   Yours (DUY) message for the response-body part; the OPES processor
   may use this information to optimize its data preservation behavior
   even before it makes the decision to preserve data.

   Example:
        Preservation-Interest-Body: 0

                                 Figure 7

3.2.7.  Content-Encodings



   A callout server MAY send a Content-Encodings list to indicate its
   preferences in content encodings.  Encodings listed first are
   preferred to other encodings.  An OPES processor MAY use any content
   encoding when sending application messages to a callout server.

   The list of preferred content encodings does not imply lack of
   support for other encodings.  The OPES processor MUST NOT bypass a
   service just because the actual content encoding does not match the
   service's preferences.

   If an OCP agent receives an application message that it cannot handle
   due to specific content encoding, the usual transaction termination
   rules apply.

   content-coding: extends atom;
   content-codings: extends list of content-coding;

   Example:
       Content-Encodings: (gzip)

                                 Figure 8

   The semantics of content-coding is defined in section 3.5 of
   [RFC2616].














Rousskov & Stecher          Standards Track                    [Page 11]

RFC 4236               HTTP Adaptation with OPES           November 2005


3.2.8.  Profile Negotiation Example



   Example:
     P: NO ({"54:http://www.iana.org/assignments/opes/ocp/http/response"
        Aux-Parts: (request-header,request-body)
        })
        SG: 5
        ;
     S: NR {"54:http://www.iana.org/assignments/opes/ocp/http/response"
        Aux-Parts: (request-header)
        Pause-At-Body: 30
        Preservation-Interest-Body: 0
        Content-Encodings: (gzip)
        }
        SG: 5
        ;

                                 Figure 9

   This example shows a negotiation offer made by an OPES processor for
   a service group (id 5) that has already been created; the callout
   server sends an adequate negotiation response.

   The OPES processor offers one profile feature for HTTP response
   messages.  Besides the standard message parts, the OPES processor is
   able to add the header and body of the original HTTP request as
   auxiliary message parts.

   The callout server requests the auxiliary request-header part, but is
   not interested in receiving the request-body part.

   The OPES processor sends at most the following message parts, in the
   specified order, for all transactions in service group 5: request-
   header, response-header, response-body, response-trailer.  Note that
   the request-body part is not included (because it is an auxiliary
   part that was not explicitly requested).  Some of the response parts
   may not be sent if the original message lacks them.

   The callout server indicates through the Preservation-Interest-Body
   parameter with size zero that it will not send any DUY messages.  The
   OPES processor may therefore preserve no preservation for any
   transaction of this profile.

   By sending a Pause-At-Body value of 30, the callout server requests a
   data pause.  The OPES processor sends a Paused My Data (DPM) message
   immediately after sending at least 30 OCTETs of the response-body
   part.  Thereafter, the OPES processor waits for a Want More Data
   (DWM) message from the callout service.



Rousskov & Stecher          Standards Track                    [Page 12]

RFC 4236               HTTP Adaptation with OPES           November 2005


3.3.  Application Message Start Message



   A new named parameter for Application Message Start (AMS) messages is
   introduced.

   AM-EL: size

                                 Figure 10

   AM-EL value is the size of the request-body part in the HTTP request
   profile, and is the size of the response-body part in the HTTP
   response profile, before any transfer codings have been applied (or
   after all transfer codings have been removed).  This definition is
   consistent with the HTTP entity length definition.

   An OCP agent that knows the exact length of the HTTP message entity
   (see Section 7.2.2 "Entity Length" in [RFC2616]) at the time it sends
   the AMS message, SHOULD announce this length using the AM-EL named
   parameter of an AMS message.  If the exact entity length is not
   known, an OCP agent MUST NOT send an AM-EL parameter.  Relaying
   correct entity length can have significant performance advantages for
   the recipient, and implementations are strongly encouraged to relay
   known entity lengths.  Similarly, relaying incorrect entity length
   can have drastic correctness consequences for the recipient, and
   implementations are urged to exercise great care when relaying entity
   length.

   An OPES processor receiving an AM-EL parameter SHOULD use the
   parameter's value in a Content-Length HTTP entity header when
   constructing an HTTP message, provided a Content-Length HTTP entity
   header is allowed for the given application message by HTTP (see
   Section 3.8.1).

3.4.  DUM Message



   A new named parameter for Data Use Mine (DUM) messages is introduced.

   AM-Part: am-part

                                 Figure 11

   An OCP agent MUST send an AM-Part parameter with every DUM message
   that is a part of an OCP transaction with an HTTP profile.  The AM-
   Part parameter value is a single am-part token.  As implied by the
   syntax, a DUM message can only contain data of a single application
   message part.  One message part can be fragmented into any number of
   DUM messages with the same AM-Part parameter.




Rousskov & Stecher          Standards Track                    [Page 13]

RFC 4236               HTTP Adaptation with OPES           November 2005


   The following example shows three DUM messages containing an abridged
   HTTP response message.  The response-body part is fragmented and sent
   within two DUM messages.

   Example:
       P: DUM 88 1 0
          Kept: 0
          AM-Part: response-header

          64:HTTP/1.1 200 OK
          Content-Type: text/html
          Content-Length: 51

          ;
       P: DUM 88 1 64
          Kept: 64
          AM-Part: response-body

          19:<html><body>This is
          ;
       P: DUM 88 1 83
          Kept: 83
          AM-Part: response-body

          32: a simple message.</body></html>
          ;



                                    Figure 12

3.5.  Selective Adaptation



   The HTTP profile for OCP applies to all HTTP messages.  That scope
   includes HTTP messages such as 1xx (Informational) responses, POST,
   CONNECT, and OPTIONS requests, as well as responses with extension
   status codes and requests with extension methods.  Unless
   specifically configured to do otherwise, an OPES processor MUST
   forward all HTTP messages for adaptation at callout servers.  OPES
   bypass instructions, configured HTTP message handling rules, and
   OCP-negotiation with a callout server are all examples of an
   acceptable "specific configuration" that provides an exception to
   this rule.

   While it may seem useless to attempt to adapt "control" messages such
   as a 100 (Continue) response, skipping such messages by default may
   lead to serious security flaws and interoperability problems.  For
   example, sensitive company information might be relayed via a





Rousskov & Stecher          Standards Track                    [Page 14]

RFC 4236               HTTP Adaptation with OPES           November 2005


   carefully crafted 100 Continue response; or a malicious CONNECT
   request may not get logged if OPES processor does not forward these
   messages to a callout service that is supposed to handle them.

   By design, OPES processor implementation cannot unilaterally decide
   that an HTTP message is not worth adapting.  It needs a callout
   server opinion, a configuration setting, or another external
   information to make the decision.

3.6.  Hop-by-hop Headers



   HTTP defines several hop-by-hop headers (e.g., Connection) and allows
   for extension headers to be specified as hop-by-hop ones (via the
   Connection header mechanism).  Depending on the environment and
   configuration, an OPES processor MAY forward hop-by-hop headers to
   callout servers and MAY use hop-by-hop headers returned by callout
   servers to build an HTTP message for the next application hop.
   However, see Section 3.7 for requirements specific to the Transfer-
   Encoding header.

   For example, a logging or statistics collection service may want to
   see hop-by-hop headers sent by the previous application hop to the
   OPES processor and/or hop-by-hop headers sent by the OPES processor
   to the next application hop.  Another service may actually handle
   HTTP logic of removing and adding hop-by-hop headers.  Many services
   will ignore hop-by-hop headers.  This specification does not define a
   mechanism for distinguishing these use cases.

3.7.  Transfer Encodings



   HTTP messages may use transfer encodings, a hop-by-hop encoding
   feature of HTTP.  Adaptations that use HTTP transfer encodings have
   to be explicitly negotiated.  This specification does not document
   such negotiations.  In the absence of explicit transfer-encoding
   negotiations, an OCP agent MUST NOT send transfer-encoded application
   message bodies.

   Informally, the above rule means that the agent or its environment
   have to make sure that all transfer encodings are stripped from an
   HTTP message body before it enters OCP scope.  An agent MUST
   terminate the OCP transaction if it has to send an application
   message body but cannot remove all transfer encodings.  Violations of
   these rules lead to interoperability problems.

   If an OCP agent receives transfer-encoded application data in
   violation of the above requirement, the agent MAY terminate the
   corresponding OCP transaction.




Rousskov & Stecher          Standards Track                    [Page 15]

RFC 4236               HTTP Adaptation with OPES           November 2005


   An OPES processor removing transfer encodings MUST remove the
   Transfer-Encoding header before sending the header part to the
   callout service.  A callout server receiving a Transfer-Encoding
   header MAY assume that original application data is still transfer-
   encoded (and terminate the transaction).  The OPES processor MUST
   send a correct Transfer-Encoding header to the next HTTP recipient,
   independent of what header (if any) the callout server returned.

   Logging and wiretapping are the examples where negotiating acceptable
   transfer encodings may be worthwhile.  While a callout server may not
   be able to strip an encoding, it may still want to log the entire
   message "as is".  In most cases, however, the callout server would
   not be able to meaningfully handle unknown transfer encodings.

3.8.  HTTP Header Correctness



   When communicating with HTTP applications, OPES processors MUST
   ensure correctness of all computable HTTP headers documented in
   specifications that the processors intend to be compliant with.  A
   computable header is defined as a header whose value can be computed
   based on the message body alone.  For example, the correctness of
   Content-Length and Content-MD5 headers has to be ensured by
   processors claiming compliance with HTTP/1.1 ([RFC2616]).

   Informally and by default, the OPES processor has to validate and
   eventually recalculate, add, or remove computable HTTP headers in
   order to build a compliant HTTP message from an adapted application
   message returned by the callout server.  If a particular OPES
   processor trusts certain HTTP headers that a callout service sends,
   it can use those headers "as is".

   An OPES processor MAY forward a partially adapted HTTP message from a
   callout server to the next callout server, without verifying HTTP
   header correctness.  Consequently, a callout service cannot assume
   that the HTTP headers it receives are correct or final from an HTTP
   point of view.

   The following subsections present guidelines for the recalculation of
   some HTTP headers.

3.8.1.  Message Size Recalculation



   By default, an OCP agent MUST NOT trust the Content-Length header
   that is sent within an HTTP header message part.  The message length
   could be modified by a callout service without adaptation of the HTTP
   message headers.





Rousskov & Stecher          Standards Track                    [Page 16]

RFC 4236               HTTP Adaptation with OPES           November 2005


   Before sending the HTTP message to the HTTP peer, the OPES processor
   has to ensure correctness of the message length indication according
   to section 4.4 of [RFC2616].

   Besides ensuring HTTP message correctness, good OPES processors set
   up the message to optimize performance, including minimizing delivery
   latency.  Specifically, indicating the end of a message by closing
   the HTTP connection ought to be the last resort:

   o  If the callout server sends an AM-EL parameter with its AMS
      message, the OPES processor SHOULD use this value to create a
      Content-Length header to be able to keep a persistent HTTP
      connection.  Note that HTTP rules prohibit a Content-Length header
      to be used in transfer-encoded messages.

   o  If AM-EL parameter or equivalent entity length information is not
      available, and HTTP rules allow for chunked transfer encoding, the
      OPES processor SHOULD use chunked transfer encoding.  Note that
      any Content-Length header has to be removed in this case.

   o  If the message size is not known a priori and chunked transfer
      coding cannot be used, but the OPES processor can wait for the OCP
      transaction to finish before forwarding the adapted HTTP message
      on a persistent HTTP connection, then the processor SHOULD compute
      and add a Content-Length header.

   o  Finally, if all optimizations are not applicable, the OPES
      processor SHOULD delete any Content-Length header and forward
      adapted data immediately, while indicating the message end by
      closing the HTTP connection.

3.8.2.  Content-MD5 Header



   By default, the OPES processor MUST assume that the callout service
   modifies the content in a way that the MD5 checksum of the message
   body becomes invalid.

   According to section 14.15 of [RFC2616], HTTP intermediaries must not
   generate Content-MD5 headers.  A recalculation is therefore possible
   only if the OPES processor is considered authoritative for the entity
   being adapted.  An un-authoritative OPES processor MUST remove the
   Content-MD5 header unless it detects that the HTTP message was not
   modified; in this case, it MAY leave the Content-MD5 header in the
   message.  When such detection significantly increases message
   latency, deleting the Content-MD5 header may be a better option.






Rousskov & Stecher          Standards Track                    [Page 17]

RFC 4236               HTTP Adaptation with OPES           November 2005


3.9.  Examples



   This is a possible OCP message flow using an HTTP request profile.
   An end-user wants to access the home page of
   www.restricted.example.com, through the proxy, but access is denied
   by a URL blocking service running on the callout server used by the
   proxy.

   OCP messages from the OPES processor are marked with "P:" and OCP
   messages from the callout server are marked with "S:".  The OCP
   connection is not closed at the end but kept open for the next OCP
   transaction.

   Example:
    P: CS;
    S: CS;
    P: SGC 11 ({"31:ocp-test.example.com/url-filter"});
    P: NO ({"53:http://www.iana.org/assignments/opes/ocp/http/request"})
       SG: 11
       ;
    S: NR {"53:http://www.iana.org/assignments/opes/ocp/http/request"}
       SG: 11
       ;
    P: TS 55 11;
    P: AMS 55
       AM-EL: 0
       ;
    P: DUM 55 0
       Kept: 0
       AM-Part: request-header
       235:GET http://www.restricted.example.com/ HTTP/1.1
       Accept: */*
       Accept-Language: de
       Accept-Encoding: gzip, deflate
       User-Agent: Mozilla/4.0 (compatible; Windows NT 5.0)
       Host: www.restricted.example.com
       Proxy-Connection: Keep-Alive


       ;
    P: AME 55;
    S: AMS 55;
    S: DUM 55 0
       AM-Part: response-header







Rousskov & Stecher          Standards Track                    [Page 18]

RFC 4236               HTTP Adaptation with OPES           November 2005


       76:HTTP/1.1 403 Forbidden
       Content-Type: text/html
       Proxy-Connection: close

       ;
    S: DUM 55 0
       AM-Part: response-body

       67:<html><body>You are not allowed to
       access this page.</body></html>
       ;
    S: AME 55;
    P: TE 55;
    S: TE 55;

                                 Figure 13

   The next example is a language translation of a small plain text file
   that gets transferred in an HTTP response.  In this example, OCP
   agents negotiate a profile for the whole OCP connection.  The OCP
   connection remains open in the end of the OCP transaction.  (Note
   that NO and NR messages were rendered with an extra new line to
   satisfy RFC formatting requirements.)

   Example:
    P: CS;
    S: CS;
    P: NO
       ({"54:http://www.iana.org/assignments/opes/ocp/http/response"});
    S: NR
       {"54:http://www.iana.org/assignments/opes/ocp/http/response"};
    P: SGC 12 ({"44:ocp-test.example.com/translate?from=EN&to=DE"});
    P: TS 89 12;
    P: AMS 89
       AM-EL: 86
       ;
    P: DUM 89 0
       AM-Part: response-header

       65:HTTP/1.1 200 OK
       Content-Type: text/plain
       Content-Length: 86


       ;
    P: DUM 89 65
       AM-Part: response-body




Rousskov & Stecher          Standards Track                    [Page 19]

RFC 4236               HTTP Adaptation with OPES           November 2005


       86:Whether 'tis nobler in the mind to suffer
       The slings and arrows of outrageous fortune
       ;
    P: AME 89;
    S: AMS 89
       AM-EL: 78
       ;
    P: TE 89;
    S: DUM 89 0
       AM-Part: response-header

       65:HTTP/1.1 200 OK
       Content-Type: text/plain
       Content-Length: 78

       ;
    S: DUM 89 63
       AM-Part: response-body

       80:Ob's edler im Gemuet, die Pfeil und Schleudern
       des wuetenden Geschicks erdulden
       ;
    S: AME 89;
    S: TE 89;

                                 Figure 14

   The following example shows modification of an HTML resource and
   demonstrates data preservation optimization.  The callout server uses
   a DUY message to send back an unchanged response header part, but
   because it does not know the size of the altered HTML resource at the
   time it sends the AMS message, the callout server omits the AM-EL
   parameter; the OPES processor is responsible for adjusting the
   Content-Length header.

   Example:
    P: CS;
    S: CS;
    P: SGC 10 ({"30:ocp-test.example.com/ad-filter"});
    P: NO ({"54:http://www.iana.org/assignments/opes/ocp/http/response"
       Aux-Parts: (request-header,request-body)
       },{"45:http://www.iana.org/assignments/opes/ocp/MIME"})
       SG: 10
       ;
    S: NR {"54:http://www.iana.org/assignments/opes/ocp/http/response"
       Aux-Parts: (request-header)
       Content-Encodings: (gzip)
       }



Rousskov & Stecher          Standards Track                    [Page 20]

RFC 4236               HTTP Adaptation with OPES           November 2005


       SG: 10
       ;
    P: TS 88 10;
    P: AMS 88
       AM-EL: 95
       ;
    P: DUM 88 0
       AM-Part: request-header

       65:GET /opes/adsample.html HTTP/1.1
       Host: www.martin-stecher.de


       ;
    P: DUM 88 65

       Kept: 65 64
       AM-Part: response-header

       64:HTTP/1.1 200 OK
       Content-Type: text/html
       Content-Length: 95


       ;
    P: DUM 88 129
       Kept: 65 90
       AM-Part: response-body

       26:<html>
       <body>
       This is my
       ;
    S: AMS 88;
    P: DUM 88 155
       Kept: 65 158
       AM-Part: response-body

       68: new ad: <img src="my_ad.gif"
       width=88 height=31>
       </body>
       </html>
       ;
    S: DUY 88 65 64
    S: DPI 88 129 2147483647;
    P: AME 88;
    S: DUM 88 0
       AM-Part: response-body



Rousskov & Stecher          Standards Track                    [Page 21]

RFC 4236               HTTP Adaptation with OPES           November 2005


       52:<html>
       <body>
       This is my new ad:
       </body>
       </html>
       ;
    S: DPI 88 129 0;
    P: TE 88;
    S: AME 88;
    S: TE 88;

                                 Figure 15

4.  Tracing



   [RFC3897] defines application-agnostic tracing facilities in OPES.
   Compliance with this specification requires compliance with
   [RFC3897].  When adapting HTTP, trace entries are supplied using HTTP
   message headers.  The following HTTP extension headers are defined to
   carry trace entries.  Their definitions are given using BNF notation
   and elements defined in [RFC2616].

        OPES-System = "OPES-System" ":" #trace-entry
        OPES-Via    = "OPES-Via" ":" #trace-entry

        trace-entry = opes-agent-id *( ";" parameter )
        opes-agent-id = absoluteURI

                                   Figure 16

   An OPES System MUST add its trace entry to the OPES-System header.
   Other OPES agents MUST use the OPES-Via header if they add their
   tracing entries.  All OPES agents MUST append their entries.
   Informally, OPES-System is the only required OPES tracing header
   while OPES-Via provides optional tracing details; both headers
   reflect the order of trace entry additions.

   If an OPES-Via header is used in the original application message, an
   OPES System MUST append its entry to the OPES-Via header.  Otherwise,
   an OPES System MAY append its entry to the OPES-Via header.  If an
   OPES System is using both headers, it MUST add identical trace
   entries except it MAY omit some or all trace-entry parameters from
   the OPES-Via header.  Informally, the OPES System entries in the
   OPES-Via header are used to delimit and group OPES-Via entries from
   different OPES Systems without having a priory knowledge about OPES
   System identifiers.





Rousskov & Stecher          Standards Track                    [Page 22]

RFC 4236               HTTP Adaptation with OPES           November 2005


   Note that all of these headers are defined using #list constructs
   and, hence, a valid HTTP message may contain multiple trace entries
   per header.  OPES agents SHOULD use a single header-field rather than
   using multiple equally-named fields to record a long trace.  Using
   multiple equally-named extension header-fields is illegal from HTTP's
   point of view and may not work with some of the OPES-unaware HTTP
   proxies.

   For example, here is an HTTP response message header after OPES
   adaptations have been applied by a single OPES processor executing 10
   OPES services:

   Example:
    HTTP/1.1 200 OK
    Date: Thu, 18 Sep 2003 06:25:24 GMT
    Last-Modified: Wed, 17 Sep 2003 18:24:25 GMT
    Content-type: application/octet-stream
    OPES-System: http://www.cdn.example.com/opes?session=ac79a749f56
    OPES-Via: http://www.cdn.example.com/opes?session=ac79a749f56,
        http://www.srvcs-4u.example.com/cat/?sid=123,
        http://www.srvcs-4u.example.com/cat/?sid=124,
        http://www.srvcs-4u.example.com/cat/?sid=125 ; mode=A

                                 Figure 17

   In the above example, the OPES processor has not included its trace
   entry or its trace entry was replaced by an OPES system trace entry.
   Only 3 out of 10 services are traced.  The remaining services did not
   include their entries or their entries were removed by OPES system or
   processor.  The last traced service included a "mode" parameter.
   Various identifiers in trace entries will probably have no meaning to
   the recipient of the message, but may be decoded by OPES System
   software.

   OPES entities MAY place optional tracing entries in a message trailer
   (i.e., entity-headers at the end of a Chunked-Body of a chunked-
   encoded message), provided trailer presence does not violate HTTP
   protocol.  See [RFC3897] for a definition of what tracing entries are
   optional.  OPES entities MUST NOT place required tracing entries in a
   message trailer.











Rousskov & Stecher          Standards Track                    [Page 23]

RFC 4236               HTTP Adaptation with OPES           November 2005


5.  Bypass



   An HTTP extension header is introduced to allow for OPES system
   bypass as defined in [RFC3897].

    OPES-Bypass  = "OPES-Bypass" ":" ( "*" | 1#bypass-entry )
    bypass-entry = opes-agent-id

                                 Figure 18

   This header can be added to HTTP requests to request OPES system
   bypass for the listed OPES agents.  The asterisk "*" character is
   used to represent all possible OPES agents.

   See [RFC3897] for what can be bypassed and for bypass requirements.

6.  IAB Considerations



   OPES treatment of IETF Internet Architecture Board (IAB)
   considerations [RFC3238] are documented in "OPES Treatment of IAB
   Considerations" [RFC3914].

7.  Security Considerations



   Application-independent security considerations are documented in
   application-agnostic OPES specifications.  HTTP profiles do not
   introduce any HTTP-specific security considerations.  However, that
   does not imply that HTTP adaptations are immune from security
   threats.

   Specific threat examples include such adaptations as rewriting the
   Request-URI of an HTTP CONNECT request or removing an HTTP hop-by-hop
   Upgrade header before the HTTP proxy can act on it.  As with any
   adaptation, the OPES agents MUST NOT perform such actions without
   HTTP client or server consent.

8.  IANA Considerations



   The IANA registers request and response profile features (Section
   3.2) using the registration procedure outlined in the "IANA
   Considerations" Section of OCP Core [RFC4037].  The corresponding
   "uri" parameters for the two features are:

   o  http://www.iana.org/assignments/opes/ocp/http/request

   o  http://www.iana.org/assignments/opes/ocp/http/response





Rousskov & Stecher          Standards Track                    [Page 24]

RFC 4236               HTTP Adaptation with OPES           November 2005


9.  Compliance



   Compliance with OPES mechanisms is defined in corresponding
   application-agnostic specifications.  HTTP profiles for these
   mechanisms use corresponding compliance definitions from these
   specifications, as if each profile were incorporated into the
   application-agnostic specification it profiles.

10.  References



10.1.  Normative References



   [RFC2616]  Fielding,  R., Gettys, J., Mogul, J., Frystyk, H.,
              Masinter, L., Leach, P., and T. Berners-Lee, "Hypertext
              Transfer Protocol -- HTTP/1.1", RFC 2616, June 1999.

   [RFC3897]  Barbir, A., "Open Pluggable Edge Services (OPES) Entities
              and End Points Communication", RFC 3897, September 2004.

   [RFC4037]  Rousskov, A., "Open Pluggable Edge Services (OPES) Callout
              Protocol (OCP) Core", RFC 4037, March 2005.

10.2.  Informative References



   [RFC3835]  Barbir, A., Penno, R., Chen, R., Hofmann, M., and H.
              Orman, "An Architecture for Open Pluggable Edge Services
              (OPES)", RFC 3835, August 2004.

   [RFC3836]  Beck, A., Hofmann, M., Orman, H., Penno, R., and A.
              Terzis, "Requirements for Open Pluggable Edge Services
              (OPES) Callout Protocols", RFC 3836, August 2004.

   [RFC3837]  Barbir, A., Batuner, O., Srinivas, B., Hofmann, M., and H.
              Orman, "Security Threats and Risks for Open Pluggable Edge
              Services (OPES)", RFC 3837, August 2004.

   [RFC3752]  Barbir, A., Burger, E., Chen, R., McHenry, S., Orman, H.,
              and R. Penno, "Open Pluggable Edge Services (OPES) Use
              Cases and Deployment Scenarios", RFC 3752, April 2004.

   [RFC3838]  Barbir, A., Batuner, O., Beck, A., Chan, T., and H. Orman,
              "Policy, Authorization, and Enforcement Requirements of
              the Open Pluggable Edge Services (OPES)", RFC 3838, August
              2004.

   [rules-p]  Beck, A. and A. Rousskov, "P: Message Processing
              Language", work in progress, October 2003.




Rousskov & Stecher          Standards Track                    [Page 25]

RFC 4236               HTTP Adaptation with OPES           November 2005


   [RFC3914]  Barbir, A. and A. Rousskov, "Open Pluggable Edge Services
              (OPES) Treatment of IAB Considerations", RFC 3914, October
              2004.

   [RFC3238]  Floyd, S. and L. Daigle, "IAB Architectural and Policy
              Considerations for Open Pluggable Edge Services", RFC
              3238, January 2002.

Acknowledgements



   The authors gratefully acknowledge the contributions of Robert
   Collins (Syncretize) and Larry Masinter (Adobe).  Larry Masinter
   provided an early review of this document.

Authors' Addresses



   Alex Rousskov
   The Measurement Factory

   EMail: rousskov@measurement-factory.com
   URI:   http://www.measurement-factory.com/


   Martin Stecher
   CyberGuard Corporation
   Vattmannstr. 3
   Paderborn  33100
   DE

   EMail: martin.stecher@webwasher.com
   URI:   http://www.webwasher.com/




















Rousskov & Stecher          Standards Track                    [Page 26]

RFC 4236               HTTP Adaptation with OPES           November 2005


Full Copyright Statement



   Copyright (C) The Internet Society (2005).

   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 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.

Acknowledgement



   Funding for the RFC Editor function is currently provided by the
   Internet Society.







Rousskov & Stecher          Standards Track                    [Page 27]