PKIX Working Group                               Michael Myers(VeriSign)
draft-ietf-pkix-ocsp-02.txt                         Rich Ankney (CertCo)
Expires in 6 months                                        February 1998


                   Internet Public Key Infrastructure
                Online Certificate Status Protocol - OCSP
                       <draft-ietf-pkix-ocsp-02.txt>



1.  Status of this Memo

This document is an Internet-Draft. Internet-Drafts  are  working
documents  of the Internet Engineering Task Force (IETF), its areas, and
its working groups. Note that other groups MAY also distribute working
documents as Internet-Drafts.

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."

To learn the current status of  any  Internet-Draft,  please  check  the
"1id-abstracts.txt"  listing  contained  in  the  Internet-Drafts Shadow
Directories   on   ftp.is.co.za   (Africa),   nic.nordu.net    (Europe),
munnari.oz.au   (Pacific  Rim),  ds.internic.net  (US  East  Coast),  or
ftp.isi.edu (US West Coast).

1. Abstract

The protocol conventions described in this document satisfy some of the
operational requirements of the Internet Public Key Infrastructure
(PKI).  This document specifies an HTTP-based application protocol
useful in determining the current status of a digital certificate
without the use of CRLs.  Additional mechanisms addressing PKIX
operational requirements are specified in separate documents.

Please send comments on this document to the ietf-pkix@tandem.com mail
list.

2. Protocol Overview

In lieu of or as a supplement to checking against a periodic CRL, it MAY
be necessary to obtain timely status regarding a  certificate’s
revocation state (cf. PKIX Part 1, Section 3.3). Examples include high-
value funds transfer or the compromise of a highly sensitive key.

The Online Certificate Status Protocol (OCSP) enables applications to
efficiently and rapidly determine the validity and revocation state of
an identified certificate.  An OCSP client issues a status request to an
OCSP responder and suspends acceptance of the subject certificate until
the responder provides a response.

The OCSP protocol is intended to be a light-weight HTTP application that
allows certificate recipients (users or application programs) to assure

Myers, Ankey                                                    [Page 1]


INTERNET DRAFT                                             February 1998

themselves, in a low-overhead manner, that a certificate is valid.  By
design, OCSP resembles HTTP in syntax and, where they overlap, in
semantics.   An anticipated goal of this decision is to leverage the
growing tools and infrastructure of the Web (most notably HTTP caching)
wherever possible.  Section 4.6 discusses the use (and limitations) of
HTTP 1.0 and HTTP 1.1 in more detail.

2.1 Request

An OCSP request contains the following data:

- protocol version
- service request
- target certificate identifier or a single end-entity certificate
- optional extensions which MAY be processed by the OCSP Responder
- an optional signature computed over the previous four fields
- additional (optional) extensions, which are not included in the
  signature computation

Upon receipt of a request, an OCSP Responder first determines if: 1) the
message is well formed, 2) the responder is configured to provide the
requested service, and 3) the responder can perform the requested
service for the subject certificate.  If any one of the prior conditions
are not met, an error message is produced; otherwise, a definitive
response is returned.

2.2 Response

All definitive response messages SHALL be digitally signed.  The key
used to sign definitive responses need not be the same signing key used
to sign the certificate. Note that caching signed responses for
frequently requested certificates MAY optionally provide some support
for reducing the cryptographic and bandwidth loads on the responder.

A definitive response message is composed of:

- response validity interval
- target certificate identifier
- certificate status value
- identification of public key needed to validate the signature
- signature algorithm OID
- optional extensions
- signature computed across hash of previous six values
- additional (optional) extensions, which are not included in the
  signature computation

This specification defines the following definitive response indicators
for use in the certificate status value:

- VALID
- INVALID              {includes reason text}
- REVOKED              {includes date of revocation}
- EXPIRED              {includes date of expiration}
- ON HOLD              {includes date of suspension}
- TRY LATER            {service temporarily unavailable}
Myers, Ankey                                                    [Page 2]


INTERNET DRAFT                                             February 1998

The path validation logic implied by the VALID and INVALID indicators is
that defined by PKIX Part 1.

The INVALID state is distinguished from the REVOKED and EXPIRED states
in that a valid certificate MAY be revoked or expired but such
information on an invalid certificate is misleading.

The ON HOLD state corresponds to valid certificates that are
operationally suspended in accordance with PKIX Part 1.

In the event that the OCSP responder is operational, but unable to
return a status for the requested certificate, the TRY LATER response
can be used to indicate that the service exists, but is temporarily
unable to respond.

Signed error messages extend the set of definitive response indicators
to include the following error conditions:

- ILLFORMED MESSAGE
- NO SERVICE
- CERTIFICATE DATA REQUIRED

A server produces the ILLFORMED MESSAGE response if the request received
does not conform to the OCSP syntax.

The response NO SERVICE MAY be produced in any circumstance in which the
server has received a well formed OCSP request but is unable to process
it.

The response CERTIFICATE DATA REQUIRED is used in cases where the server
requires the client to supply the certificate data itself in order to
construct a response.

2.3 Response Pre-production

The response validity interval noted in the prior section is composed of
a {produced-at, next-update} pair of elements in the response syntax.
Section 4.2 provides details of the response syntax.

OCSP responders MAY pre-produce signed responses reflecting the current
status of certificates at the time the response was produced.  The time
at which the response was produced SHALL be reflected in the produced-at
field of the response.

The producer of the response MAY include a value for next-update.  The
exact interval between produced-at and next-update for given response is
a matter of local security and operational policy.  If the next-update
field is not present, the response is valid only for the particular
request which prompted it.  Equivalently, the next-update field is
considered to be the same as the produced-at field.

If responses are pre-produced, then for a given certificate, the
periodicity of this pre-production SHOULD match the response validity
interval of the most recently produced response.


Myers, Ankey                                                    [Page 3]


INTERNET DRAFT                                             February 1998

3. Functional Requirements

3.1 Certificate Content

In order to convey to OCSP clients a well-known point of information
access, CAs SHALL provide the capability to include the
AuthorityInfoAccess extension (defined in PKIX Part 1, section 4.2.2.1)
in certificates intended to be applied to the service.

CAs that support an OCSP service, either hosted locally or provided by a
Trusted Third Party, SHALL provide a value for a
uniformResourceIndicator (URI) accessLocation and the OID value id-pkix-
ocsp for the accessMethod in the AccessDescription SEQUENCE.

The value of the accessLocation field in the subject certificate
corresponds to the URL placed into an OCSP request (see section 5.1).

3.2 Request Generation and Submission

OCSP clients SHALL be capable of transmitting OCSP as an HTTP 1.1 GET
and of receiving the response as the Entity-Body of an HTTP 1.1 Full-
Response.  OCSP Clients MAY be capable of transmitting OCSP as an HTTP
1.1 POST with the certificate being queried forming the message body.
OCSP servers SHALL support HTTP 1.1 GET requests and MAY support HTTP
1.1 POST requests. Section 4.5 discusses use of HTTP transport.

3.3 Error Responses

Upon receipt of a request which fails to parse, the receiving OCSP
responder SHALL respond with an error message.  If the responder is
configured to provide signed error responses, a failure to parse an
incoming request SHALL be indicated by an ILLFORMED MESSAGE response.
The value of the identifier of such a response SHALL be NULL_ID.

For service requests not supported by the responder, the responder SHALL
respond with an error message. If the responder is configured to provide
signed error responses, non-availability of the requested service SHALL
be indicated by a NO SERVICE response.

This protocol makes use of HTTP as a transport.  OCSP clients SHOULD
consequently enable automatic recovery from a lost connection.  An HTTP
timeout mechanism is one conventional means of doing so.

3.4 Status Responses

Upon receipt of an OCSP request containing an end-entity certificate, if
the certificate fails to validate against Section 6 of PKIX Part 1 for
reasons other than revocation, OCSP esponders SHALL respond with
INVALID.  Responses MAY be supplemented with explanatory text that
provides additional context.  Section 5.2 of this document specifies a
minimal set of explanatory text for this purpose.




Myers, Ankey                                                    [Page 4]


INTERNET DRAFT                                             February 1998

3.5 Signed Response Acceptance Requirements

Prior to accepting a signed response as valid, OCSP clients SHALL
confirm that:

1.  The certificate identified in a received response corresponds to
that which was identified in a former request;

2.  The signature on the response is valid;

3.  The identity of the signer matches the intended recipient of the
request.

4. Detailed Protocol

The protocol consists of an OCSP request and an OCSP response.  The
request is conveyed as part of an HTTP GET or POST request, and its
syntax is compatible with that defined in [URL].  In particular,
reserved characters in the request are encoded using the %xx mechanism
of [URL], unless they are used for their reserved purposes (in this
case, delimiting fields in the request).  Note that this encoding
requirement includes the three special characters (plus sign, forward
slash, and equal sign) used in the base64 encoding mechanism. Components
in the request are thus of the form:

token           = *(ALPHA / DIGIT / SAFE / EXTRA)
safe        = "$" | "-" | "_" | "." | "+"
extra       = "!" | "*" | "'" | "(" | ")" | ","

Components (and sub-fields within the components) are delimited using
several of the reserved characters:

reserved    = ";" | "/" | "?" | ":" | "@" | "&" | "="

The following (punctuation) characters are always encoded, as they might
have meaning outside the scope of the URL.  E.g., the pound sign (#) is
used to separate a URL and a fragment identifier.

punctuation     = "<" | ">" | "#" | "%" | <">

Requests and responses are composed of:

  -     fixed fields (required in all messages), followed by
  -     optional fields, which MUST be supported by an OCSP responder,
        but need not be present in all messages, and
  -     optional extensions, which MAY be supported by an OCSP responder.

Optional fields and extensions use a simple name/value syntax.

Definitive responses and definitive error responses MUST be signed.
Requests MAY be signed using the request-signature extension.
Signatures are computed from the beginning of the request or response
through the first two fields of the signature block.  Extensions MAY be
present before the signature (in which case they are included in the

Myers, Ankey                                                    [Page 5]


INTERNET DRAFT                                             February 1998

signature computation), or after the signature (in which case they are
not signed.

4.1 Request Syntax

An OCSP request is an HTTP 1.1 GET or POST method composed of a URL
followed by a sequence of required and optional fields. The following
grammar specifies the request portion of the protocol.  Quoted syntactic
elements are terminal elements of the grammar, which is the ABNF grammar
from [ABNF].

OCSP-request     =      url "/ocsp/ver/1/" target [extensions]
[signature] [extensions]
url              =      protocol “://” domain-name “/”
protocol         =      “http”
time             =      <date and time, as: YYYYMMDDHHMMSSZ>
target           =      cert-id
cert-id          =      “ID” “/” hash
hash             =      <sha1-hash(Issuer DN | cert serial number)>
extensions           =  *extension
extension            =  “/” ext-name “/” ext-value
ext-name             =  ALPHA  *(ALPHA / DIGIT / “-“)
ext-value            =  token
DQUOTE       =  %x22
signature            =  “/” “sig” “/” signature-block
signature-block  =      key-id “&” sig-alg-oid “&” signature-value
key-id           =      <subject key ID of certificate needed to
validate signature, per PKIX-1; defaults to
SHA-1 hash of public key in certificate>
sig-alg-oid      =      <algorithm combination used to produce sig,
                                   as dotted integer e.g. 1.2.3.4>
signature-value  =    <base-64 encoded value corresponding to
                           the result of using sig-alg-oid>

The value of “2” for the version field accommodates preliminary
implementations of a different request and response syntax.

To produce a value for the cert-id field, the client first calculates a
SHA-1 hash across the concatenation of Issuer DN with the serial number
in the target certificate, base-64 encodes the hash and appends the
result to the prior fields.

The HTTP POST method is to be used in cases where the certificate data
itself is required by the server. The body of the request has content
type application/xxxx and consists of the certificate data itself. Since
HTTP requires connections to be ‘8-bit clean’ no additional encoding of
the certificate data is required.

Support for extensions is OPTIONAL.  This standard defines several
useful extensions in Section 4.5.  Additional extensions MAY be defined
in additional RFCs.  All extensions use the name/value syntax described
above.  Unrecognized extensions can be ignored, by skipping past the
slash delimiting the value, then skipping the value.


Myers, Ankey                                                    [Page 6]


INTERNET DRAFT                                             February 1998

The requester signature is used to authenticate the requester to the
OCSP Responder.  It is used in conjunction with the requester
certificate extension defined below.

The signature is computed over the portion of the request which precedes
the signature, i.e. the URL, request, and target fields, any extensions
which precede the signature, and the key-id and sig-alg-oid portions of
this extension.  Note that extensions following this one are not
included; typically the requester certificate extension does not need to
be signed, and would be positioned after the signature. Thus, the
signature is computed over:

{ url request target extensions sig-fields }

The OCSP Responder verifies the signature, using the public key
identified by key-id and the requester certificate extension defined
below.

4.2 Response Syntax

An HTTP-based OCSP response is composed of a sequence of data fields
similar to the OCSP request.  Response codes are returned as the ASCII
encoding of a decimal number.  Values with a minus sign (ASCII encoding
of “-”) indicate definitive error values.

OCSP-response      =    definitive-rsp / error-rsp
definitive-rsp     =    base status-value [extensions] signature
[extensions]
error-rsp          =    minimal-error / definitive-error
minimal-error      =    %x20                             ; “ “
definitive-error   =    base error-value [extensions] signature
                                [extensions]

base               =    produced-at “/” prior-id “/” next-update
produced-at        =    time
next-update         =   “next-update” “/” time
prior-id           =    <cert-id of prior request>
error-value        =    illformed-msg / no-service /certificate-reqd
illformed-msg      =    %x2d %x31                       ; “-1”
no-service         =    %x2d %x32                       ; “-2”
certificate-reqd   =    %x2d %x33                       ; “-3”
status-value       =    status-code [“/” reason-text ]
[ “/” date-context ]
status-code        =    valid /invalid / revoked / not-revoked /
expired / try-later
valid              =    %x31                            ; “1”
invalid            =    %x32                            ; “2”
revoked            =    %x33                            ; “3”
expired            =    %x34                            ; “4”
on-hold            =    %x35                            ; “5”
try-later                =      %x36                                      ; “6”
reason-text        =    “reason” “/” token    ; for additional context
date-context       =    “on” “/” time


Myers, Ankey                                                    [Page 7]


INTERNET DRAFT                                             February 1998

Standard values for reason-text SHALL include:
“1   The root for this certificate is not trusted on this responder.”
“2   Could not find CA’s public key.”
“3   CA’s public key invalid.”
“4   CA’s public key revoked.”
“5   CA’s public key expired.”
“6   CA not authorized for Subject’s name.”
“7   CA not authorized for Subject’s privileges.”
“8   CA’s public key did not validate signature.”
“9   Could not find CA’s revocation information.”
“10  CA’s CRL out of date.”

When producing REVOKED or HOLD responses, OCSP responders SHALL include
the date of the revocation in the status-value field as a value for
date-context.

The produced-at and next-update fields define a validity interval.  This
interval corresponds to the {thisUpdate, nextUpdate} CRL validity
interval.  Responses whose next-update value is earlier than the local
system time value SHOULD be considered unreliable.

To produce a value for the cert-id field, the client first calculates an
MD5 hash across the concatenation of Issuer DN with the serial number
in the target certificate, base-64 encodes the hash and appends the
result to the prior fields.

To produce a signed response, the responder first calculates a hash of
the sequence of characters from the beginning of the response through
the “&” following the sig-alg-oid subfield of the signature block.
Using the above syntax, this is the sequence:

        base status-value [extensions] sig-fields

where

        sig-fields = “/” “sig” “/” key-id “&” sig-alg-oid “&”

The responder signs the hash, base-64 encodes the result and then
appends it to the prior fields, encoding any special characters in the
base64 signature as described in [URL].  The associated hash and signing
algorithms are identified by the value of sig-alg-oid.

4.3 Mandatory and Optional Cryptographic Algorithms

Clients that request OCSP services SHALL be capable of processing
responses signed used DSA keys identified by the DSA sig-alg-oid
specified in section 7.2.2 of PKIX Part 1.  Clients SHOULD also be
capable of processing RSA signatures as specified in ection 7.2.1 of
PKIX Part 1.

4.4  Extensions

This section defines some standard extensions.  Support for all
extensions is OPTIONAL.  For each extension, the definition indicates

Myers, Ankey                                                    [Page 8]


INTERNET DRAFT                                             February 1998

its syntax, processing performed by the OCSP Responder, and any
extensions which are included in the corresponding response.  Note that
extensions use a simple name/value syntax.

4.4.1  Nonce

The nonce cryptographically binds a request and response, to prevent
replay attacks.  The nonce extension has syntax:

        “nonce” “/” atom

and MAY be any non-repeating value (e.g., random number, transaction
hash, counter, or timestamp).  The OCSP Responder returns the nonce as a
response extension.  The nonce is included in the response signature
computation.  The nonce structure is opaque to the OCSP Responder.

4.4.2 Requester Certificate

This extension conveys the certificate of the signer, and is used by the
OCSP Responder to verify the requester signature.  It has syntax:

        “reqcert” “/” <base64 encoding of a certificate>
        “reqid” “/” hash                ; MD5(issuer || serial #)

The cert production is either a base64-encoded certificate, or the
hash of the issuer and serial number of the certificate as described in
Section 4.1.

NOTE:  Depending on the OCSP implementation, the key-id contained
in the key-id portion of the signature extension MAY be sufficient
to identify the key needed to verify the signature.  In this case,
this extension is not needed.

NOTE:  This extension contains a single certificate.  Other(CA)
certificates MAY be needed for the OCSP to verify the signature.
This could be included in the request, as additional instances of
this extension.  Or another extension could be defined, which
might convey a SEQUENCE OF Certificate, or an empty CMS SignedData
instance (with the certificates field populated appropriately).
Construction of the relevant certification path by the OCSP
Responder could (instead) just be declared out of scope for this
standard.

4.4.3 CRL References

It MAY be desirable for the OCSP responder to indicate the CRL on which
a revoked or held certificate is found.  This can be useful where OCSP
is used between repositories, and also as an auditing mechanism.  Three
extensions are defined for this purpose:

“crlurl” “/” token
“crlnum” “/” number
“crltime” “/” time
number = *DIGIT

Myers, Ankey                                                    [Page 9]


INTERNET DRAFT                                             February 1998

The crlurl extension contains a URL which can be used to retrieve the
CRL the requested certificate is listed on. Note that any reserved
characters in the token must be encoded per [URL].

The crlnum extension indicates the value in the CRL number extension of
the relevant CRL.

The crltime extension indicates the value in the thisUpdate field of the
relevant CRL.

4.4.4 Policy Identifier

This extension contains a policy ID and relevant qualifiers.  It MAY be
sent in a request to indicate the policy which the requester considers
acceptable.  This extension MAY also be included in the response.  The
syntax is:

        “policy” “/” <base64 encoding of a PolicyInformation structure>

4.5 Responder Key Identification

It is possible that an OCSP responder MAY have more than one valid
public signature key of the same cryptographic algorithm.  To assist
clients in identifying which public key to use, OCSP responders SHALL
include in all signed responses a SHA-1 hash of the required public key.

It is also possible that an OCSP client MAY be in possession of more
than one valid certificate containing the OCSP responder’s public key.

This specification asserts no constraints on the means by which clients
determine which certificate to use.

4.6 HTTP Transport Mechanism

The request syntax is intended to mimic an object retrieval via HTTP 1.1
in order for it to be cached by local proxy responders.

OCSP requests are composed as an HTTP GET as follows:

GET <request> HTTP/1.1
<headers>

OCSP requests are composed as an HTTP POST as follows:

POST <request> HTTP/1.1
<headers>

<certificate>

The response to such a query is the Entity-Body of an HTTP 1.1 Full-
Response as defined in RFC 2068 with Content-Type: application/ocsp.




Myers, Ankey                                                   [Page 10]


INTERNET DRAFT                                             February 1998

A representation of the HTTP context of an OCSP request and response
follows.  The content differs slightly from current request and response
syntax.

GET /status/check/ver/1/ID/qyXLklpfK2wYd8iPsGdOwQ== HTTP/1.1
Host: status.our-ca.com
Date: Thu, 23 Oct 1997 22:33:30 GMT
User-Agent: Transport xy/z


HTTP/1.0 200 OK
Server: Netscape-Enterprise/2.0a
Date: Thu, 23 Oct 1997 22:33:34 GMT
Last-modified: Thursday, 23 Oct 1997 12:38:25 GMT
Expires: Thursday, 23 Oct 1997 23:38:25 GMT
Content-type: application/octet-stream

19971023123825Z&qyXLklpfK2wYd8iPsGdOwQ==&2&1.3.14.3.2.15

#uFom3GIAHjIdlWZ5SsFKTvGXHgML35n21zsQvFGT3hWmULBsvH6MDg4+FY55P6wgwxAWTSV
S3h8xFiacN9m5S4xBDO/5IpVFpFwdhrSe8S5/jYK2qPGsGdjzCmGQIX03CaGLh+NOn8x9Wpo
wtnCMhg4UeDZm+b4BKrmNpT6g0Mw=

In cases where the server returns the response CERTIFICATE DATA REQUIRED
the request must be repeated using the POST method to provide the server
with the certificate data. Note that base64 encoding is specified in the
following example for readability alone. A production client would
normally use binary encoding.

POST /status/check/ver/1/ID/qyXLklpfK2wYd8iPsGdOwQ== HTTP/1.1
Host: status.our-ca.com
Date: Thu, 23 Oct 1997 22:33:30 GMT
User-Agent: Transport xy/z
Content-Type: application/X509
Content-Encoding: Base-64

2+3094fh9p2831410t493upoiqwjf9p832174rtawoirtf0913274r5hqefpq0w7rfqolaiw
yerq32047r9qwherfpqw40/r7qp9w84utrp92q34723q940+3q4tp0u3q4tu3qp40t73q04t
7r0q394ut0qwu4t0[9qw40t7q3/048u3q4wtu3q0+4t0+3q47034q7t093749734097t0709
3475t023q743wq/890++ewrtpoiausrpoiurpsoituqwpoet/0==

HTTP/1.0 200 OK
Server: Netscape-Enterprise/2.0a
Date: Thu, 23 Oct 1997 22:33:34 GMT
Last-modified: Thursday, 23 Oct 1997 12:38:25 GMT
Expires: Thursday, 23 Oct 1997 23:38:25 GMT
Content-type: application/octet-stream

19971023123825Z&qyXLklpfK2wYd8iPsGdOwQ==&2&1.3.14.3.2.15

#uFom3GIAHjIdlWZ5SsFKTvGXHgML35n21zsQvFGT3hWmULBsvH6MDg4+FY55P6wgwxAWTSV
S3h8xFiacN9m5S4xBDO/5IpVFpFwdhrSe8S5/jYK2qPGsGdjzCmGQIX03CaGLh+NOn8x9Wpo
wtnCMhg4UeDZm+b4BKrmNpT6g0Mw=


Myers, Ankey                                                   [Page 11]


INTERNET DRAFT                                             February 1998

If an HTTP/1.0 server receives the message or a proxy downgrades the
request to a 1.0 message the request will work as expected except that
the connection will be closed at the end of the request, a keep-alive
request will not be supported. The advantages of using HTTP 1.1 over
HTTP 1.0 are:

1) HTTP 1.0 is a Best Commercial Practices document while HTTP 1.1 is
full standards track.

2) It would be possible to send a sequence of requests at the same time
using the 'keep-alive' facility. This would allow the raw data for
validating a certificate to be collected in a single server transaction
or allow a whole 'address book' of certificates to be checked in one go.

3) The 1.1 cache control features would be available. These allow
clients to specify precisely the degree of staleness they will permit.
They can also make statements like 'get me the latest if you can get it
otherwise send me something no more than 240 seconds old.'

3a) A particularly useful feature is that the server can require caches
to poll to check freshness each time they serve a piece of data. In the
HTTP/1.0 model there is only the pragma: no-cache which prohibits all
caching entirely.

5. Security Considerations

For this service to be effective, certificate using systems must connect
to the certificate status service provider. In the event such a
connection cannot be obtained, certificate-using systems could implement
CRL processing logic as a fall-back position.

A denial of service vulnerability is evident with respect to a flood of
queries constructed to produce error responses.  The production of a
cryptographic signature significantly affects response generation cycle
time, thereby exacerbating the situation. Performance studies on a
preliminary implementation of OCSP capable of handling two million hits
per day without degradation suggest this effect is of an orders of
magnitude per response. Unsigned error responses provide a reasonable
tradeoff against protection against this particular attack.

The use of unsigned error messages introduces a vulnerability to
intermediation attacks. It is reasonable to ask for error messages to be
signed to address this vulnerability.  A request to do so however must
also consider the converse risk identified above—namely that increasing
the response cycle time of error messages through use of cryptographic
signing increases the impact of flooding attacks.  Parties implementing
OCSP responders that wish to offer the benefit of signed error responses
SHOULD strongly consider the use of hardware-assisted cryptography.
Doing so will reduce the threat of flood attacks.

The use of precomputed responses MAY allow replay attacks in which an
old (VALID) response is replayed prior to its expiration date but after
the certificate has been revoked.  HTTP caching MAY also allow replay of

Myers, Ankey                                                   [Page 12]

INTERNET DRAFT                                             February 1998

stale responses; see Section 4.5 for a discussion of how caching MAY be
controlled in HTTP 1.1.

6. References

[HTTP] Hypertext Transfer Protocol -- HTTP/1.0. T. Berners-Lee,
       R. Fielding & H. Frystyk, RFC 1945, May 1996.

[ABNF] Augmented BNF for Syntax Specifications: ABNF.  D. Crocker,
P.  Overell, RFC 2234, November 1997.

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

[URL] Uniform Resource Locators (URL), T. Berners-Lee, L. Masinter,
M.  McCahill, RFC 1738, December 1994.

7. Author’s Address

Michael Myers
VeriSign, Inc.
1390 Shorebird Way
Mountain View, CA 94019
mmyers@verisign.com

Rich Ankney
CertCo, LLC
13506 King Charles Dr.
Chantilly, VA  20151
rankney@erols.com

Appendix A

Registration of OCSP as a MIME type

To: IANA@isi.edu
Subject: Registration of new MIME content-type/subtype

MIME type name: application

MIME subtype name: ocsp

Required parameters: none

Optional parameters: none

Encoding considerations: none

Security considerations:

This data information which may be conceivably be used to enforce legal contracts,
resolve disputes or transfer financial risk from one party to another.  The design
of software that processes this type should be trustworthy in its design and
operations.

Published specification:  This document