HTTP代理协议(HTTP CONNECT)

Abstract

 

  This memo explains how to usethe Upgrade mechanism in HTTP/1.1 to
   initiate Transport LayerSecurity (TLS) over an existing TCP
   connection. This allowsunsecured and secured HTTP traffic to share
   the same well known port (inthis case, http: at 80 rather than
   https: at 443). It alsoenables "virtual hosting", so a single HTTP +
   TLS server can disambiguatetraffic intended for several hostnames at
   a single IP address.

   Since HTTP/1.1 [1] definesUpgrade as a hop-by-hop mechanism, this
   memo also documents the HTTPCONNECT method for establishing end-to-
   end tunnels across HTTPproxies. Finally, this memo establishes new
   IANA registries for publicHTTP status codes, as well as public or
   private Upgrade producttokens.

   This memo does NOT affect thecurrent definition of the 'https' URI
   scheme, which already definesa separate namespace
   (http://example.org/ andhttps://example.org/ are not equivalent).

1. Motivation
   The historical practice ofdeploying HTTP over SSL3 [3] has
   distinguished the combinationfrom HTTP alone by a unique URI scheme
   and the TCP port number. Thescheme 'http' meant the HTTP protocol
   alone on port 80, while'https' meant the HTTP protocol over SSL on
   port 443. Parallel well-known port numbers have similarly been
   requested -- and in somecases, granted -- to distinguish between
   secured and unsecured use ofother application protocols (e.g.
   snews, ftps). This approacheffectively halves the number of
   available well knownports.

   At the Washington DC IETFmeeting in December 1997, the Applications
   Area Directors and the IESGreaffirmed that the practice of issuing
   parallel "secure" port numbersshould be deprecated. The HTTP/1.1
   Upgrade mechanism can applyTransport Layer Security [6] to an open
   HTTP connection.

   In the nearly two years since,there has been broad acceptance of the
   concept behind this proposal,but little interest in implementing
   alternatives to port 443 forgeneric Web browsing. In fact, nothing
   in this memo affects thecurrent interpretation of https: URIs.
   However, new applicationprotocols built atop HTTP, such as the
   Internet Printing Protocol[7], call for just such a mechanism in
   order to move ahead in theIETF standards process.

   The Upgrade mechanism alsosolves the "virtual hosting" problem.
   Rather than allocatingmultiple IP addresses to a single host, an
   HTTP/1.1 server will use theHost: header to disambiguate the
   intended web service. AsHTTP/1.1 usage has grown more prevalent,
   more ISPs are offeringname-based virtual hosting, thus delaying IP
   address spaceexhaustion.

   TLS (and SSL) have beenhobbled by the same limitation as earlier
   versions of HTTP: the initialhandshake does not specify the intended
   hostname, relying exclusivelyon the IP address. Using a cleartext
   HTTP/1.1 Upgrade: preamble tothe TLS handshake -- choosing the
   certificates based on theinitial Host: header -- will allow ISPs to
   provide secure name-basedvirtual hosting as well.

2. Introduction

   TLS, a.k.a., SSL (SecureSockets Layer), establishes a private end-
   to-end connection, optionallyincluding strong mutual authentication,
   using a variety ofcryptosystems. Initially, a handshake phase uses
   three subprotocols to set up arecord layer, authenticate endpoints,
   set parameters, as well asreport errors.  Then, there is an ongoing
   layered record protocol thathandles encryption, compression, and
   reassembly for the remainderof the connection. The latter is
   intended to be completelytransparent. For example, there is no
   dependency between TLS'srecord markers and or certificates and
   HTTP/1.1's chunked encoding orauthentication.

   Either the client or servercan use the HTTP/1.1 [1] Upgrade
   mechanism (Section 14.42) toindicate that a TLS-secured connection
   is desired or necessary. Thismemo defines the "TLS/1.0" Upgrade
   token, and a new HTTP StatusCode, "426 Upgrade Required".

   Section 3 and Section 4describe the operation of a directly
   connected client and server.Intermediate proxies must establish an
   end-to-end tunnel beforeapplying those operations, as explained in
   Section 5.

3. Client Requested Upgrade to HTTP over TLS

   When the client sends anHTTP/1.1 request with an Upgrade header
   field containing the token"TLS/1.0", it is requesting the server to
   complete the current HTTP/1.1request after switching to TLS/1.0.

3.1 Optional Upgrade

   A client MAY offer to switchto secured operation during any clear
   HTTP request when an unsecuredresponse would be acceptable:

      GET http://example.bank.com/acct_stat.html?749394889300HTTP/1.1
      Host: example.bank.com
      Upgrade: TLS/1.0
      Connection: Upgrade

   In this case, the server MAYrespond to the clear HTTP operation
   normally, OR switch to securedoperation (as detailed in the next
   section).

   Note that HTTP/1.1 [1]specifies "the upgrade keyword MUST be
   supplied within a Connectionheader field (section 14.10) whenever
   Upgrade is present in anHTTP/1.1 message".

3.2 Mandatory Upgrade

   If an unsecured response wouldbe unacceptable, a client MUST send an
   OPTIONS request first tocomplete the switch to TLS/1.0 (if
   possible).

      OPTIONS * HTTP/1.1
      Host: example.bank.com
      Upgrade: TLS/1.0
      Connection: Upgrade

3.3 Server Acceptance of Upgrade Request

   As specified in HTTP/1.1 [1],if the server is prepared to initiate
   the TLS handshake, it MUSTsend the intermediate "101 Switching
   Protocol" and MUST include anUpgrade response header specifying the
   tokens of the protocol stackit is switching to:

      HTTP/1.1 101 Switching Protocols
      Upgrade: TLS/1.0, HTTP/1.1
      Connection: Upgrade

   Note that the protocol tokenslisted in the Upgrade header of a 101
   Switching Protocols responsespecify an ordered 'bottom-up' stack.

   As specifiedin  HTTP/1.1 [1], Section 10.1.2: "The serverwill
   switch protocols to thosedefined by the response's Upgrade header
   field immediately after theempty line which terminates the 101
   response".

   Once the TLS handshakecompletes successfully, the server MUST
   continue with the response tothe original request. Any TLS handshake
   failure MUST lead todisconnection, per the TLS error alert
   specification.

4. Server Requested Upgrade to HTTP over TLS

   The Upgrade response headerfield advertises possible protocol
   upgrades a server MAY accept.In conjunction with the "426 Upgrade
   Required" status code, aserver can advertise the exact protocol
   upgrade(s) that a client MUSTaccept to complete the request.

4.1 Optional Advertisement

   As specified in HTTP/1.1 [1],the server MAY include an Upgrade
   header in any response otherthan 101 or 426 to indicate a
   willingness to switch to any(combination) of the protocols listed.

4.2 Mandatory Advertisement

   A server MAY indicate that aclient request can not be completed
   without TLS using the "426Upgrade Required" status code, which MUST
   include an an Upgrade headerfield specifying the token of the
   required TLS version.

      HTTP/1.1 426 Upgrade Required
      Upgrade: TLS/1.0, HTTP/1.1
      Connection: Upgrade

   The server SHOULD include amessage body in the 426 response which
   indicates in human readableform the reason for the error and
   describes any alternativecourses which may be available to the user.

   Note that even if a client iswilling to use TLS, it must use the
   operations in Section 3 toproceed; the TLS handshake cannot begin
   immediately after the 426response.

5. Upgrade across Proxies

   As a hop-by-hop header,Upgrade is negotiated between each pair of
   HTTPcounterparties.  If a User Agent sends a requestwith an Upgrade
   header to a proxy, it isrequesting a change to the protocol between
   itself and the proxy, not anend-to-end change.

   Since TLS, in particular,requires end-to-end connectivity to provide
   authentication and preventman-in-the-middle attacks, this memo
   specifies the CONNECT methodto establish a tunnel across proxies.

   Once a tunnel is established,any of the operations in Section 3 can
   be used to establish a TLSconnection.

5.1 Implications of Hop By Hop Upgrade
   If an origin server receivesan Upgrade header from a proxy and
   responds with a 101 SwitchingProtocols response, it is changing the
   protocol only on theconnection between the proxy and itself.
   Similarly, a proxy mightreturn a 101 response to its client to
   change the protocol on thatconnection independently of the protocols
   it is using to communicatetoward the origin server.

   These scenarios alsocomplicate diagnosis of a 426 response. Since
   Upgrade is a hop-by-hopheader, a proxy that does not recognize 426
   might remove the accompanyingUpgrade header and prevent the client
   from determining the requiredprotocol switch.  If a client receives
   a 426 status without anaccompanying Upgrade header, it will need to
   request an end to end tunnelconnection as described in Section 5.2
   and repeat the request inorder to obtain the required upgrade
   information.

   This hop-by-hop definition ofUpgrade was a deliberate choice.  It
   allows for incrementaldeployment on either side of proxies, and for
   optimized protocols betweencascaded proxies without the knowledge of
   the parties that are not apart of the change.

5.2 Requesting a Tunnel with CONNECT
   A CONNECT method requests thata proxy establish a tunnel connection
   on its behalf. The Request-URIportion of the Request-Line is always
   an 'authority' as defined byURI Generic Syntax [2], which is to say
   the host name and port numberdestination of the requested connection
   separated by a colon:

     CONNECT server.example.com:80 HTTP/1.1
     Host: server.example.com:80

   Other HTTP mechanisms can beused normally with the CONNECT method --
   except end-to-end protocolUpgrade requests, of course, since the
   tunnel must be establishedfirst.

   For example, proxyauthentication might be used to establish the
   authority to create atunnel:

     CONNECT server.example.com:80 HTTP/1.1
     Host: server.example.com:80
     Proxy-Authorization: basic aGVsbG86d29ybGQ=

   Like any other pipelinedHTTP/1.1 request, data to be tunneled may be
   sent immediately after theblank line. The usual caveats also apply:
   data may be discarded if theeventual response is negative, and the
   connection may be reset withno response if more than one TCP segment
   is outstanding.

5.3 Establishing a Tunnel with CONNECT

   Any successful (2xx) responseto a CONNECT request indicates that the
   proxy has established aconnection to the requested host and port,
   and has switched to tunnelingthe current connection to that server
   connection.

   It may be the case that theproxy itself can only reach the requested
   origin server through anotherproxy.  In this case, the first proxy
   SHOULD make a CONNECT requestof that next proxy, requesting a tunnel
   to theauthority.  A proxy MUST NOT respond with any 2xxstatus code
   unless it has either a director tunnel connection established to the
   authority.

   An origin server whichreceives a CONNECT request for itself MAY
   respond with a 2xx status codeto indicate that a connection is
   established.

   If at any point either one ofthe peers gets disconnected, any
   outstanding data that camefrom that peer will be passed to the other
   one, and after that also theother connection will be terminated by
   the proxy. If there isoutstanding data to that peer undelivered,
   that data will bediscarded.

6. Rationale for the use of a 4xx (client error) Status Code
   Reliable, interoperablenegotiation of Upgrade features requires an
   unambiguous failure signal.The 426 Upgrade Required status code
   allows a server todefinitively state the precise protocol extensions
   a given resource must beserved with.

   It might at first appear thatthe response should have been some form
   of redirection (a 3xx code),by analogy to an old-style redirection
   to an https:URI.  User agents that do not understandUpgrade:
   preclude this.

   Suppose that a 3xx code hadbeen assigned for "Upgrade Required"; a
   user agent that did notrecognize it would treat it as 300.  Itwould
   then properly look for a"Location" header in the response and
   attempt to repeat the requestat the URL in that header field. Since
   it did not know to Upgrade toincorporate the TLS layer, it would at
   best fail again at the newURL.

7.1 HTTP Status Code Registry
   The HTTP Status Code Registrydefines the name space for the Status-
   Code token in the Status lineof an HTTP response.  The initial
   values for this name space arethose specified by:

   1.  DraftStandard for HTTP/1.1 [1]
   2.  WebDistributed Authoring and Versioning [4] [defines 420-424]
   3.  WebDAVAdvanced Collections [5] (Work in Progress) [defines 425]
   4.  Section 6[defines 426]

   Values to be added to thisname space SHOULD be subject to review in
   the form of a standards trackdocument within the IETF Applications
   Area.  Anysuch document SHOULD be traceable through statuses of
   either 'Obsoletes' or'Updates' to the Draft Standard for
   HTTP/1.1 [1].

7.2 HTTP Upgrade Token Registry
   The HTTP Upgrade TokenRegistry defines the name space for product
   tokens used to identifyprotocols in the Upgrade HTTP header field.
   Each registered token shouldbe associated with one or a set of
   specifications, and withcontact information.

   The Draft Standard forHTTP/1.1 [1] specifies that these tokens obey
   the production for'product':
     product        = token ["/" product-version]
     product-version = token
   Registrations should beallowed on a First Come First Served basis as
   described in BCP 26 [10].These specifications need not be IETF
   documents or be subject toIESG review, but should obey the following
   rules:

   1.  A token,once registered, stays registered forever.
   2.  Theregistration MUST name a responsible party for the
      registration.
   3.  Theregistration MUST name a point of contact.
   4.  Theregistration MAY name the documentation required for the
      token.
   5.  Theresponsible party MAY change the registration at any time.
      The IANA will keep a record of all such changes, and makethem
      available upon request.
   6.  Theresponsible party for the first registration of a "product"
      token MUST approve later registrations of a "version" token
      together with that "product" token before they can beregistered.
   7.  Ifabsolutely required, the IESG MAY reassign the responsibility
      for a token. This will normally only be used in the case whena
      responsible party cannot be contacted.

   This specification defines theprotocol token "TLS/1.0" as the
   identifier for the protocolspecified by The TLS Protocol [6].

   It is NOT required thatspecifications for upgrade tokens be made
   publicly available, but thecontact information for the registration
   SHOULD be.