4 NETWORK WORKING GROUP A. Menon-Sen
5 Internet-Draft Oryx Mail Systems GmbH
6 Intended status: Standards Track A. Melnikov
7 Expires: February 1, 2010 Isode Ltd
14 Salted Challenge Response (SCRAM) SASL Mechanism
15 draft-ietf-sasl-scram-04.txt
19 This Internet-Draft is submitted to IETF in full conformance with the
20 provisions of BCP 78 and BCP 79.
22 Internet-Drafts are working documents of the Internet Engineering
23 Task Force (IETF), its areas, and its working groups. Note that
24 other groups may also distribute working documents as Internet-
27 Internet-Drafts are draft documents valid for a maximum of six months
28 and may be updated, replaced, or obsoleted by other documents at any
29 time. It is inappropriate to use Internet-Drafts as reference
30 material or to cite them other than as "work in progress."
32 The list of current Internet-Drafts can be accessed at
33 http://www.ietf.org/ietf/1id-abstracts.txt.
35 The list of Internet-Draft Shadow Directories can be accessed at
36 http://www.ietf.org/shadow.html.
38 This Internet-Draft will expire on February 1, 2010.
42 Copyright (c) 2009 IETF Trust and the persons identified as the
43 document authors. All rights reserved.
45 This document is subject to BCP 78 and the IETF Trust's Legal
46 Provisions Relating to IETF Documents in effect on the date of
47 publication of this document (http://trustee.ietf.org/license-info).
48 Please review these documents carefully, as they describe your rights
49 and restrictions with respect to this document.
55 Menon-Sen, et al. Expires February 1, 2010 [Page 1]
57 Internet-Draft SCRAM July 2009
62 The secure authentication mechanism most widely deployed and used by
63 Internet application protocols is the transmission of clear-text
64 passwords over a channel protected by Transport Layer Security (TLS).
65 There are some significant security concerns with that mechanism,
66 which could be addressed by the use of a challenge response
67 authentication mechanism protected by TLS. Unfortunately, the
68 challenge response mechanisms presently on the standards track all
69 fail to meet requirements necessary for widespread deployment, and
70 have had success only in limited use.
72 This specification describes a family of Simple Authentication and
73 Security Layer (SASL, RFC 4422) authentication mechanisms called the
74 Salted Challenge Response Authentication Mechanism (SCRAM), which
75 addresses the security concerns and meets the deployability
76 requirements. When used in combination with TLS or an equivalent
77 security layer, a mechanism from this family could improve the
78 status-quo for application protocol authentication and provide a
79 suitable choice for a mandatory-to-implement mechanism for future
80 application protocol standards.
111 Menon-Sen, et al. Expires February 1, 2010 [Page 2]
113 Internet-Draft SCRAM July 2009
118 1. Conventions Used in This Document . . . . . . . . . . 4
119 1.1. Terminology . . . . . . . . . . . . . . . . . . . . . 4
120 1.2. Notation . . . . . . . . . . . . . . . . . . . . . . . 5
121 2. Introduction . . . . . . . . . . . . . . . . . . . . . 7
122 3. SCRAM Algorithm Overview . . . . . . . . . . . . . . . 9
123 4. SCRAM Mechanism Names . . . . . . . . . . . . . . . . 10
124 5. SCRAM Authentication Exchange . . . . . . . . . . . . 11
125 5.1. SCRAM Attributes . . . . . . . . . . . . . . . . . . . 12
126 6. Channel Binding . . . . . . . . . . . . . . . . . . . 15
127 6.1. Default Channel Binding . . . . . . . . . . . . . . . 16
128 7. Formal Syntax . . . . . . . . . . . . . . . . . . . . 17
129 8. SCRAM as a GSS-API Mechanism . . . . . . . . . . . . . 20
130 8.1. GSS-API Principal Name Types for SCRAM . . . . . . . . 20
131 8.2. GSS-API Per-Message Tokens for SCRAM . . . . . . . . . 20
132 8.3. GSS_Pseudo_random() for SCRAM . . . . . . . . . . . . 21
133 9. Security Considerations . . . . . . . . . . . . . . . 22
134 10. IANA Considerations . . . . . . . . . . . . . . . . . 24
135 11. Acknowledgements . . . . . . . . . . . . . . . . . . . 26
136 Appendix A. Other Authentication Mechanisms . . . . . . . . . . . 27
137 Appendix B. Design Motivations . . . . . . . . . . . . . . . . . . 28
138 Appendix C. Internet-Draft Change History . . . . . . . . . . . . 29
139 12. References . . . . . . . . . . . . . . . . . . . . . . 31
140 12.1. Normative References . . . . . . . . . . . . . . . . . 31
141 12.2. Normative References for GSS-API implementors . . . . 31
142 12.3. Informative References . . . . . . . . . . . . . . . . 32
143 Authors' Addresses . . . . . . . . . . . . . . . . . . 34
167 Menon-Sen, et al. Expires February 1, 2010 [Page 3]
169 Internet-Draft SCRAM July 2009
172 1. Conventions Used in This Document
174 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
175 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
176 document are to be interpreted as described in [RFC2119].
178 Formal syntax is defined by [RFC5234] including the core rules
179 defined in Appendix B of [RFC5234].
181 Example lines prefaced by "C:" are sent by the client and ones
182 prefaced by "S:" by the server. If a single "C:" or "S:" label
183 applies to multiple lines, then the line breaks between those lines
184 are for editorial clarity only, and are not part of the actual
189 This document uses several terms defined in [RFC4949] ("Internet
190 Security Glossary") including the following: authentication,
191 authentication exchange, authentication information, brute force,
192 challenge-response, cryptographic hash function, dictionary attack,
193 eavesdropping, hash result, keyed hash, man-in-the-middle, nonce,
194 one-way encryption function, password, replay attack and salt.
195 Readers not familiar with these terms should use that glossary as a
198 Some clarifications and additional definitions follow:
200 o Authentication information: Information used to verify an identity
201 claimed by a SCRAM client. The authentication information for a
202 SCRAM identity consists of salt, iteration count, the "StoredKey"
203 and "ServerKey" (as defined in the algorithm overview) for each
204 supported cryptographic hash function.
206 o Authentication database: The database used to look up the
207 authentication information associated with a particular identity.
208 For application protocols, LDAPv3 (see [RFC4510]) is frequently
209 used as the authentication database. For network-level protocols
210 such as PPP or 802.11x, the use of RADIUS is more common.
212 o Base64: An encoding mechanism defined in [RFC4648] which converts
213 an octet string input to a textual output string which can be
214 easily displayed to a human. The use of base64 in SCRAM is
215 restricted to the canonical form with no whitespace.
217 o Octet: An 8-bit byte.
219 o Octet string: A sequence of 8-bit bytes.
223 Menon-Sen, et al. Expires February 1, 2010 [Page 4]
225 Internet-Draft SCRAM July 2009
228 o Salt: A random octet string that is combined with a password
229 before applying a one-way encryption function. This value is used
230 to protect passwords that are stored in an authentication
235 The pseudocode description of the algorithm uses the following
238 o ":=": The variable on the left hand side represents the octet
239 string resulting from the expression on the right hand side.
241 o "+": Octet string concatenation.
243 o "[ ]": A portion of an expression enclosed in "[" and "]" may not
244 be included in the result under some circumstances. See the
245 associated text for a description of those circumstances.
247 o HMAC(key, str): Apply the HMAC keyed hash algorithm (defined in
248 [RFC2104]) using the octet string represented by "key" as the key
249 and the octet string "str" as the input string. The size of the
250 result is the hash result size for the hash function in use. For
251 example, it is 20 octets for SHA-1 (see [RFC3174]).
253 o H(str): Apply the cryptographic hash function to the octet string
254 "str", producing an octet string as a result. The size of the
255 result depends on the hash result size for the hash function in
258 o XOR: Apply the exclusive-or operation to combine the octet string
259 on the left of this operator with the octet string on the right of
260 this operator. The length of the output and each of the two
261 inputs will be the same for this use.
267 U0 := HMAC(str, salt + INT(1))
271 Ui-1 := HMAC(str, Ui-2)
272 Ui := HMAC(str, Ui-1)
274 Hi := U0 XOR U1 XOR U2 XOR ... XOR Ui
279 Menon-Sen, et al. Expires February 1, 2010 [Page 5]
281 Internet-Draft SCRAM July 2009
284 where "i" is the iteration count, "+" is the string concatenation
285 operator and INT(g) is a four-octet encoding of the integer g,
286 most significant octet first.
288 o This is, essentially, PBKDF2 [RFC2898] with HMAC() as the PRF and
289 with dkLen == output length of HMAC() == output length of H().
335 Menon-Sen, et al. Expires February 1, 2010 [Page 6]
337 Internet-Draft SCRAM July 2009
342 This specification describes a family of authentication mechanisms
343 called the Salted Challenge Response Authentication Mechanism (SCRAM)
344 which addresses the requirements necessary to deploy a challenge-
345 response mechanism more widely than past attempts. When used in
346 combination with Transport Layer Security (TLS, see [RFC5246]) or an
347 equivalent security layer, a mechanism from this family could improve
348 the status-quo for application protocol authentication and provide a
349 suitable choice for a mandatory-to-implement mechanism for future
350 application protocol standards.
352 For simplicity, this family of mechanisms does not presently include
353 negotiation of a security layer [RFC4422]. It is intended to be used
354 with an external security layer such as that provided by TLS or SSH,
355 with optional channel binding [RFC5056] to the external security
358 SCRAM is specified herein as a pure Simple Authentication and
359 Security Layer (SASL) [RFC4422] mechanism, but it conforms to the new
360 bridge between SASL and the Generic Security Services Application
361 Programming Interface (GSS-API) called "GS2" [I-D.ietf-sasl-gs2].
362 This means that this document defines both, a SASL mechanism and a
365 SCRAM provides the following protocol features:
367 o The authentication information stored in the authentication
368 database is not sufficient by itself to impersonate the client.
369 The information is salted to prevent a pre-stored dictionary
370 attack if the database is stolen.
372 o The server does not gain the ability to impersonate the client to
373 other servers (with an exception for server-authorized proxies).
375 o The mechanism permits the use of a server-authorized proxy without
376 requiring that proxy to have super-user rights with the back-end
379 o Mutual authentication is supported, but only the client is named
380 (i.e., the server has no name).
382 A separate document defines a standard LDAPv3 [RFC4510] attribute
383 that enables storage of the SCRAM authentication information in LDAP.
384 See [I-D.melnikov-sasl-scram-ldap].
386 For an in-depth discussion of why other challenge response mechanisms
387 are not considered sufficient, see appendix A. For more information
391 Menon-Sen, et al. Expires February 1, 2010 [Page 7]
393 Internet-Draft SCRAM July 2009
396 about the motivations behind the design of this mechanism, see
399 Comments regarding this draft may be sent either to the
400 ietf-sasl@imc.org mailing list or to the authors.
447 Menon-Sen, et al. Expires February 1, 2010 [Page 8]
449 Internet-Draft SCRAM July 2009
452 3. SCRAM Algorithm Overview
454 Note that this section omits some details, such as client and server
455 nonces. See Section 5 for more details.
457 To begin with, the SCRAM client is in possession of a username and
458 password. It sends the username to the server, which retrieves the
459 corresponding authentication information, i.e. a salt, StoredKey,
460 ServerKey and the iteration count i. (Note that a server
461 implementation may chose to use the same iteration count for all
462 accounts.) The server sends the salt and the iteration count to the
463 client, which then computes the following values and sends a
464 ClientProof to the server:
467 SaltedPassword := Hi(password, salt)
468 ClientKey := HMAC(SaltedPassword, "Client Key")
469 StoredKey := H(ClientKey)
470 AuthMessage := client-first-message-bare + "," +
471 server-first-message + "," +
472 client-final-message-without-proof
473 ClientSignature := HMAC(StoredKey, AuthMessage)
474 ClientProof := ClientKey XOR ClientSignature
475 ServerKey := HMAC(SaltedPassword, "Server Key")
476 ServerSignature := HMAC(ServerKey, AuthMessage)
479 The server authenticates the client by computing the ClientSignature,
480 exclusive-ORing that with the ClientProof to recover the ClientKey
481 and verifying the correctness of the ClientKey by applying the hash
482 function and comparing the result to the StoredKey. If the ClientKey
483 is correct, this proves that the client has access to the user's
486 Similarly, the client authenticates the server by computing the
487 ServerSignature and comparing it to the value sent by the server. If
488 the two are equal, it proves that the server had access to the user's
491 The AuthMessage is computed by concatenating messages from the
492 authentication exchange. The format of these messages is defined in
503 Menon-Sen, et al. Expires February 1, 2010 [Page 9]
505 Internet-Draft SCRAM July 2009
508 4. SCRAM Mechanism Names
510 A SCRAM mechanism name is a string "SCRAM-" followed by the
511 uppercased name of the underlying hash function taken from the IANA
512 "Hash Function Textual Names" registry (see http://www.iana.org),
513 optionally followed by the suffix "-PLUS" (see below). Note that
514 SASL mechanism names are limited to 20 characters, which means that
515 only hash function names with lengths shorter or equal to 9
516 characters (20-length("SCRAM-")-length("-PLUS") can be used. For
517 cases when the underlying hash function name is longer than 9
518 characters, an alternative 9 character (or shorter) name can be used
519 to construct the corresponding SCRAM mechanism name, as long as this
520 alternative name doesn't conflict with any other hash function name
521 from the IANA "Hash Function Textual Names" registry.
523 For interoperability, all SCRAM clients and servers MUST implement
524 the SCRAM-SHA-1 authentication mechanism, i.e. an authentication
525 mechanism from the SCRAM family that uses the SHA-1 hash function as
526 defined in [RFC3174].
528 The "-PLUS" suffix is used only when the server supports channel
529 binding to the external channel. If the server supports channel
530 binding, it will advertise both the "bare" and "plus" versions of
531 whatever mechanisms it supports (e.g., if the server supports only
532 SCRAM with SHA-1 then it will advertise support for both SCRAM-SHA-1
533 and SCRAM-SHA-1-PLUS); if the server does not support channel
534 binding, then it will advertise only the "bare" version of the
535 mechanism (e.g., only SCRAM-SHA-1). The "-PLUS" exists to allow
536 negotiation of the use of channel binding. See Section 6.
559 Menon-Sen, et al. Expires February 1, 2010 [Page 10]
561 Internet-Draft SCRAM July 2009
564 5. SCRAM Authentication Exchange
566 SCRAM is a SASL mechanism whose client response and server challenge
567 messages are text-based messages containing one or more attribute-
568 value pairs separated by commas. Each attribute has a one-letter
569 name. The messages and their attributes are described in
570 Section 5.1, and defined in Section 7.
572 This is a simple example of a SCRAM-SHA-1 authentication exchange
573 when the client doesn't support channel bindings:
576 C: n,,n=Chris Newman,r=ClientNonce
577 S: r=ClientNonceServerNonce,s=PxR/wv+epq,i=128
578 C: c=biwsCg==,r=ClientNonceServerNonce,p=WxPv/siO5l+qxN4
582 [[anchor5: Note that the all hashes above are fake and will be fixed
585 With channel-binding data sent by the client this might look like
586 this (see [tls-server-end-point] for the definition of tls-server-
587 end-point TLS channel binding):
590 C: p=tls-server-end-point,,n=Chris Newman,r=ClientNonce
591 S: r=ClientNonceServerNonce,s=PxR/wv+epq,i=128
592 C: c=cD10bHMtc2VydmVyLWVuZC1wb2ludCwsy1hFtXOnZ+ySrQM6srFp
593 l/77uqvtxrg7nBY1BetEr/g=,r=ClientNonceServerNonce,p=Wx
598 [[anchor6: Note that all hashes above are fake and will be fixed
601 First, the client sends a message containing:
603 o a GS2 header consisting of a flag indicating whether channel
604 binding is supported-but-not-used, not supported, or used, and an
605 optional SASL authorization identity;
607 o SCRAM username and a random, unique nonce attributes.
609 Note that the client's first message will always start with "n", "y"
610 or "p", otherwise the message is invalid and authentication MUST
611 fail. This is important, as it allows for GS2 extensibility (e.g.,
615 Menon-Sen, et al. Expires February 1, 2010 [Page 11]
617 Internet-Draft SCRAM July 2009
620 to add support for security layers).
622 In response, the server sends the user's iteration count i, the
623 user's salt, and appends its own nonce to the client-specified one.
624 The client then responds with the same nonce and a ClientProof
625 computed using the selected hash function as explained earlier. The
626 server verifies the nonce and the proof, verifies that the
627 authorization identity (if supplied by the client in the first
628 message) is authorized to act as the authentication identity, and,
629 finally, it responds with a ServerSignature, concluding the
630 authentication exchange. The client then authenticates the server by
631 computing the ServerSignature and comparing it to the value sent by
632 the server. If the two are different, the client MUST consider the
633 authentication exchange to be unsuccessful and it might have to drop
636 5.1. SCRAM Attributes
638 This section describes the permissible attributes, their use, and the
639 format of their values. All attribute names are single US-ASCII
640 letters and are case-sensitive.
642 Note that the order of attributes in client or server messages is
643 fixed, with the exception of extension attributes (described by the
644 "extensions" ABNF production), which can appear in any order in the
645 designated positions. See the ABNF section for authoritative
648 o a: This is an optional attribute, and is part of the GS2
649 [I-D.ietf-sasl-gs2] bridge between the GSS-API and SASL. This
650 attribute specifies an authorization identity. A client may
651 include it in its first message to the server if it wants to
652 authenticate as one user, but subsequently act as a different
653 user. This is typically used by an administrator to perform some
654 management task on behalf of another user, or by a proxy in some
657 Upon the receipt of this value the server verifies its
658 correctness according to the used SASL protocol profile.
659 Failed verification results in failed authentication exchange.
661 If this attribute is omitted (as it normally would be), the
662 authorization identity is assumed to be derived from the
663 username specified with the (required) "n" attribute.
665 The server always authenticates the user specified by the "n"
666 attribute. If the "a" attribute specifies a different user,
667 the server associates that identity with the connection after
671 Menon-Sen, et al. Expires February 1, 2010 [Page 12]
673 Internet-Draft SCRAM July 2009
676 successful authentication and authorization checks.
678 The syntax of this field is the same as that of the "n" field
679 with respect to quoting of '=' and ','.
681 o n: This attribute specifies the name of the user whose password is
682 used for authentication (a.k.a. "authentication identity"
683 [RFC4422]). A client MUST include it in its first message to the
684 server. If the "a" attribute is not specified (which would
685 normally be the case), this username is also the identity which
686 will be associated with the connection subsequent to
687 authentication and authorization.
689 Before sending the username to the server, the client MUST
690 prepare the username using the "SASLPrep" profile [RFC4013] of
691 the "stringprep" algorithm [RFC3454]. If the preparation of
692 the username fails or results in an empty string, the client
693 SHOULD abort the authentication exchange (*).
695 (*) An interactive client can request a repeated entry of the
698 Upon receipt of the username by the server, the server SHOULD
699 prepare it using the "SASLPrep" profile [RFC4013] of the
700 "stringprep" algorithm [RFC3454]. If the preparation of the
701 username fails or results in an empty string, the server SHOULD
702 abort the authentication exchange.
704 The characters ',' or '=' in usernames are sent as '=2C' and
705 '=3D' respectively. If the server receives a username which
706 contains '=' not followed by either '2C' or '3D', then the
707 server MUST fail the authentication.
709 o m: This attribute is reserved for future extensibility. In this
710 version of SCRAM, its presence in a client or a server message
711 MUST cause authentication failure when the attribute is parsed by
714 o r: This attribute specifies a sequence of random printable
715 characters excluding ',' which forms the nonce used as input to
716 the hash function. No quoting is applied to this string. As
717 described earlier, the client supplies an initial value in its
718 first message, and the server augments that value with its own
719 nonce in its first response. It is important that this value be
720 different for each authentication. The client MUST verify that
721 the initial part of the nonce used in subsequent messages is the
722 same as the nonce it initially specified. The server MUST verify
723 that the nonce sent by the client in the second message is the
727 Menon-Sen, et al. Expires February 1, 2010 [Page 13]
729 Internet-Draft SCRAM July 2009
732 same as the one sent by the server in its first message.
734 o c: This REQUIRED attribute specifies base64-encoded of a header
735 and the channel-binding data. It is sent by the client in its
736 second authentication message. The header consist of:
738 * the GS2 header from the client's first message (recall: a
739 channel binding flag and an optional authzid). This header is
740 going to include channel binding type prefix (see [RFC5056]),
741 if and only if the client is using channel binding;
743 * followed by the external channel's channel binding data, if and
744 only if the client is using channel binding.
746 o s: This attribute specifies the base64-encoded salt used by the
747 server for this user. It is sent by the server in its first
748 message to the client.
750 o i: This attribute specifies an iteration count for the selected
751 hash function and user, and MUST be sent by the server along with
754 For SCRAM-SHA-1/SCRAM-SHA-1-PLUS SASL mechanism servers SHOULD
755 announce a hash iteration-count of at least 4096. Note that a
756 client implementation MAY cache SaltedPassword/ClientKey for
757 later reauthentication to the same service, as it is likely
758 that the server is going to advertise the same salt value upon
759 reauthentication. This might be useful for mobile clients
760 where CPU usage is a concern.
762 o p: This attribute specifies a base64-encoded ClientProof. The
763 client computes this value as described in the overview and sends
766 o v: This attribute specifies a base64-encoded ServerSignature. It
767 is sent by the server in its final message, and is used by the
768 client to verify that the server has access to the user's
769 authentication information. This value is computed as explained
783 Menon-Sen, et al. Expires February 1, 2010 [Page 14]
785 Internet-Draft SCRAM July 2009
790 SCRAM supports channel binding to external secure channels, such as
791 TLS. Clients and servers may or may not support channel binding,
792 therefore the use of channel binding is negotiable. SCRAM does not
793 provide security layers, however, therefore it is imperative that
794 SCRAM provide integrity protection for the negotiation of channel
797 Use of channel binding is negotiated as follows:
799 o Servers SHOULD advertise both non-PLUS (SCRAM-<hash-function>) and
800 the PLUS-variant (SCRAM-<hash-function>-PLUS) SASL mechanism
801 names. If the server cannot support channel binding, it MAY
802 advertise only the non-PLUS variant. If the server would never
803 succeed authentication of the non-PLUS variant due to policy
804 reasons, it MAY advertise only the PLUS-variant.
806 o If the client negotiates mechanisms then the client MUST select
807 SCRAM-<hash-function>-PLUS if offered by the server and the client
808 wants to select SCRAM with the given hash function. Otherwise
809 (the client does not negotiate mechanisms), if the client has no
810 prior knowledge about mechanisms supported by the server and
811 wasn't explicitly configured to use a particular variant of the
812 SCRAM mechanism, then it MUST select only SCRAM-<hash-function>
813 (not suffixed with "-PLUS").
815 o If the client supports channel binding and the server appears to
816 support it (i.e., the client sees SCRAM-<hash-function>-PLUS), or
817 if the client wishes to use channel binding but the client does
818 not negotiate mechanisms, then the client MUST set the GS2 channel
819 binding flag to "p" in order to indicate the channel binding type
820 it is using and it MUST include the channel binding data for the
821 external channel in the computation of the "c=" attribute (see
824 o If the client supports channel binding but the server does not
825 appear to (i.e., the client did not see SCRAM-<hash-function>-
826 PLUS) then the client MUST either fail authentication or it MUST
827 choose the non-PLUS mechanism and set the GS2 channel binding flag
828 to "y" and MUST NOT include channel binding data for the external
829 channel in the computation of the "c=" attribute (see
832 o If the client does not support channel binding then the client
833 MUST set the GS2 channel binding flag to "n" and MUST NOT include
834 channel binding data for the external channel in the computation
835 of the "c=" attribute (see Section 5.1).
839 Menon-Sen, et al. Expires February 1, 2010 [Page 15]
841 Internet-Draft SCRAM July 2009
844 o Upon receipt of the client first message the server checks the GS2
845 channel binding flag (gs2-cb-flag).
847 * If the flag is set to "y" and the server supports channel
848 binding the server MUST fail authentication. This is because
849 if the client sets the GS2 channel binding flag set to "y" then
850 the client must have believed that the server did not support
851 channel binding -- if the server did in fact support channel
852 binding then this is an indication that there has been a
853 downgrade attack (e.g., an attacker changed the server's
854 mechanism list to exclude the -PLUS suffixed SCRAM mechanism
857 * If the channel binding flag was "p" and the server does not
858 support the indicated channel binding type then the server MUST
861 The server MUST always validate the client's "c=" field. The server
862 does this by constructing the value of the "c=" attribute and then
863 checking that it matches the client's c= attribute value.
865 For more discussions of channel bindings, and the syntax of the
866 channel binding data for various security protocols, see [RFC5056].
868 6.1. Default Channel Binding
870 A default channel binding type agreement process for all SASL
871 application protocols that do not provide their own channel binding
872 type agreement is provided as follows.
874 'tls-unique' is the default channel binding type for any application
875 that doesn't specify one.
877 Servers MUST implement the "tls-unique" [tls-unique]
878 [I-D.altman-tls-channel-bindings] channel binding type, if they
879 implement any channel binding. Clients SHOULD implement the "tls-
880 unique" [tls-unique] [I-D.altman-tls-channel-bindings] channel
881 binding type, if they implement any channel binding. Clients and
882 servers SHOULD choose the highest- layer/innermost end-to-end TLS
883 channel as the channel to bind to.
885 Servers MUST choose the channel binding type indicated by the client,
886 or fail authentication if they don't support it.
895 Menon-Sen, et al. Expires February 1, 2010 [Page 16]
897 Internet-Draft SCRAM July 2009
902 The following syntax specification uses the Augmented Backus-Naur
903 Form (ABNF) notation as specified in [RFC5234]. "UTF8-2", "UTF8-3"
904 and "UTF8-4" non-terminal are defined in [RFC3629].
907 ALPHA = <as defined in RFC 5234 appendix B.1>
908 DIGIT = <as defined in RFC 5234 appendix B.1>
909 UTF8-2 = <as defined in RFC 3629 (STD 63)>
910 UTF8-3 = <as defined in RFC 3629 (STD 63)>
911 UTF8-4 = <as defined in RFC 3629 (STD 63)>
913 attr-val = ALPHA "=" value
914 ;; Generic syntax of any attribute sent
915 ;; by server or client
919 value-safe-char = %x01-2B / %x2D-3C / %x3E-7F /
920 UTF8-2 / UTF8-3 / UTF8-4
921 ;; UTF8-char except NUL, "=", and ",".
923 value-char = value-safe-char / "="
925 base64-char = ALPHA / DIGIT / "/" / "+"
927 base64-4 = 4base64-char
929 base64-3 = 3base64-char "="
931 base64-2 = 2base64-char "=="
933 base64 = *base64-4 [base64-3 / base64-2]
935 posit-number = %x31-39 *DIGIT
938 saslname = 1*(value-safe-char / "=2C" / "=3D")
939 ;; Conforms to <value>
941 authzid = "a=" saslname
942 ;; Protocol specific.
944 cb-name = 1*(ALPHA / DIGIT / "." / "-")
945 ;; See RFC 5056 section 7.
946 ;; E.g. "tls-server-end-point" or
951 Menon-Sen, et al. Expires February 1, 2010 [Page 17]
953 Internet-Draft SCRAM July 2009
956 gs2-cbind-flag = "p=" cb-name / "n" / "y"
957 ;; "n" -> client doesn't support channel binding
958 ;; "y" -> client does support channel binding
959 ;; but thinks the server does not.
960 ;; "p" -> client requires channel binding.
961 ;; The selected channel binding follows "p=".
963 gs2-header = gs2-cbind-flag "," [ authzid ] ","
964 ;; GS2 header for SCRAM
965 ;; (the actual GS2 header includes an optional
966 ;; flag to indicate that the GSS mechanism is not
967 ;; "standard" but since SCRAM is "standard" we
968 ;; don't include that flag).
970 username = "n=" saslname
971 ;; Usernames are prepared using SASLPrep.
973 reserved-mext = "m=" 1*(value-char)
974 ;; Reserved for signalling mandatory extensions.
975 ;; The exact syntax will be defined in
978 channel-binding = "c=" base64
979 ;; base64 encoding of cbind-input
983 nonce = "r=" c-nonce [s-nonce]
984 ;; Second part provided by server.
992 verifier = "v=" base64
993 ;; base-64 encoded ServerSignature.
995 iteration-count = "i=" posit-number
998 client-first-message-bare =
1000 username "," nonce ["," extensions]
1002 client-first-message =
1003 gs2-header client-first-message-bare
1007 Menon-Sen, et al. Expires February 1, 2010 [Page 18]
1009 Internet-Draft SCRAM July 2009
1012 server-first-message =
1013 [reserved-mext ","] nonce "," salt ","
1014 iteration-count ["," extensions]
1016 client-final-message-without-proof =
1017 channel-binding "," nonce [","
1020 client-final-message =
1021 client-final-message-without-proof "," proof
1023 gss-server-error = "e=" value
1024 server-final-message = gss-server-error /
1025 verifier ["," extensions]
1026 ;; The error message is only for the GSS-API
1027 ;; form of SCRAM, and it is OPTIONAL to
1030 extensions = attr-val *("," attr-val)
1031 ;; All extensions are optional,
1032 ;; i.e. unrecognized attributes
1033 ;; not defined in this document
1036 cbind-data = 1*OCTET
1038 cbind-input = gs2-header [ cbind-data ]
1039 ;; cbind-data MUST be present for
1040 ;; gs2-cbind-flag of "p" and MUST be absent
1063 Menon-Sen, et al. Expires February 1, 2010 [Page 19]
1065 Internet-Draft SCRAM July 2009
1068 8. SCRAM as a GSS-API Mechanism
1070 This section and its sub-sections and all normative references of it
1071 not referenced elsewhere in this document are INFORMATIONAL for SASL
1072 implementors, but they are NORMATIVE for GSS-API implementors.
1074 SCRAM is actually also GSS-API mechanism. The messages are the same,
1075 but a) the GS2 header on the client's first message and channel
1076 binding data is excluded when SCRAM is used as a GSS-API mechanism,
1077 and b) the RFC2743 section 3.1 initial context token header is
1078 prefixed to the client's first authentication message (context
1081 The GSS-API mechanism OID for SCRAM is <TBD> (see Section 10).
1083 8.1. GSS-API Principal Name Types for SCRAM
1085 SCRAM does not name acceptors. Therefore only GSS_C_NO_NAME and
1086 names of type GSS_C_NT_ANONYMOUS shall be allowed as the target name
1087 input of GSS_Init_sec_context() when using a SCRAM mechanism.
1089 SCRAM supports only a single name type for initiators:
1090 GSS_C_NT_USER_NAME. GSS_C_NT_USER_NAME is the default name type for
1093 There is no name canonicalization procedure for SCRAM beyond applying
1094 SASLprep as described in Section 5.1.
1096 The query, display and exported name syntax for SCRAM principal names
1097 is the same: there is no syntax -- SCRAM principal names are free-
1098 form. (The exported name token does, of course, conform to [RFC2743]
1099 section 3.2, but the "NAME" part of the token is just a SCRAM user
1102 8.2. GSS-API Per-Message Tokens for SCRAM
1104 The per-message tokens for SCRAM as a GSS-API mechanism SHALL be the
1105 same as those for the Kerberos V GSS-API mechanism [RFC4121], using
1106 the Kerberos V "aes128-cts-hmac-sha1-96" enctype [RFC3962].
1108 The 128-bit session key SHALL be derived by using the least
1109 significant (right-most) 128 bits of HMAC(StoredKey, "GSS-API session
1110 key" || ClientKey || AuthMessage).
1112 SCRAM does support PROT_READY, and is PROT_READY on the initiator
1113 side first upon receipt of the server's reply to the initial security
1119 Menon-Sen, et al. Expires February 1, 2010 [Page 20]
1121 Internet-Draft SCRAM July 2009
1124 8.3. GSS_Pseudo_random() for SCRAM
1126 The GSS_Pseudo_random() [RFC4401] for SCRAM SHALL be the same as for
1127 the Kerberos V GSS-API mechanism [RFC4402]. There is no acceptor-
1128 asserted sub-session key for SCRAM, thus GSS_C_PRF_KEY_FULL and
1129 GSS_C_PRF_KEY_PARTIAL are equivalent for SCRAM's GSS_Pseudo_random().
1175 Menon-Sen, et al. Expires February 1, 2010 [Page 21]
1177 Internet-Draft SCRAM July 2009
1180 9. Security Considerations
1182 If the authentication exchange is performed without a strong security
1183 layer, then a passive eavesdropper can gain sufficient information to
1184 mount an offline dictionary or brute-force attack which can be used
1185 to recover the user's password. The amount of time necessary for
1186 this attack depends on the cryptographic hash function selected, the
1187 strength of the password and the iteration count supplied by the
1188 server. An external security layer with strong encryption will
1189 prevent this attack.
1191 If the external security layer used to protect the SCRAM exchange
1192 uses an anonymous key exchange, then the SCRAM channel binding
1193 mechanism can be used to detect a man-in-the-middle attack on the
1194 security layer and cause the authentication to fail as a result.
1195 However, the man-in-the-middle attacker will have gained sufficient
1196 information to mount an offline dictionary or brute-force attack.
1197 For this reason, SCRAM includes the ability to increase the iteration
1200 If the authentication information is stolen from the authentication
1201 database, then an offline dictionary or brute-force attack can be
1202 used to recover the user's password. The use of salt mitigates this
1203 attack somewhat by requiring a separate attack on each password.
1204 Authentication mechanisms which protect against this attack are
1205 available (e.g., the EKE class of mechanisms). RFC 2945 [RFC2945] is
1206 an example of such technology. There are IPR disclosures at
1207 http://datatracker.ietf.org/ipr/ that mention RFC 2945.
1209 If an attacker obtains the authentication information from the
1210 authentication repository and either eavesdrops on one authentication
1211 exchange or impersonates a server, the attacker gains the ability to
1212 impersonate that user to all servers providing SCRAM access using the
1213 same hash function, password, iteration count and salt. For this
1214 reason, it is important to use randomly-generated salt values.
1216 SCRAM does not negotiate a hash function to use. Hash function
1217 negotiation is left to the SASL mechanism negotiation. It is
1218 important that clients be able to sort a locally available list of
1219 mechanisms by preference so that the client may pick the most
1220 preferred of a server's advertised mechanism list. This preference
1221 order is not specified here as it is a local matter. The preference
1222 order should include objective and subjective notions of mechanism
1223 cryptographic strength (e.g., SCRAM with a successor to SHA-1 may be
1224 preferred over SCRAM with SHA-1).
1226 Note that to protect the SASL mechanism negotiation applications
1227 normally must list the server mechs twice: once before and once after
1231 Menon-Sen, et al. Expires February 1, 2010 [Page 22]
1233 Internet-Draft SCRAM July 2009
1236 authentication, the latter using security layers. Since SCRAM does
1237 not provide security layers the only ways to protect the mechanism
1238 negotiation are: a) use channel binding to an external channel, or b)
1239 use an external channel that authenticates a user-provided server
1242 SCRAM does not protect against downgrade attacks of channel binding
1243 types. The complexities of negotiation a channel binding type, and
1244 handling down-grade attacks in that negotiation, was intentionally
1245 left out of scope for this document.
1247 A hostile server can perform a computational denial-of-service attack
1248 on clients by sending a big iteration count value.
1250 See [RFC4086] for more information about generating randomness.
1287 Menon-Sen, et al. Expires February 1, 2010 [Page 23]
1289 Internet-Draft SCRAM July 2009
1292 10. IANA Considerations
1294 IANA is requested to add the following family of SASL mechanisms to
1295 the SASL Mechanism registry established by [RFC4422]:
1299 Subject: Registration of a new SASL family SCRAM
1301 SASL mechanism name (or prefix for the family): SCRAM-*
1302 Security considerations: Section 7 of [RFCXXXX]
1303 Published specification (optional, recommended): [RFCXXXX]
1304 Person & email address to contact for further information:
1305 IETF SASL WG <ietf-sasl@imc.org>
1306 Intended usage: COMMON
1307 Owner/Change controller: IESG <iesg@ietf.org>
1308 Note: Members of this family must be explicitly registered
1309 using the "IETF Consensus" registration procedure.
1310 Reviews must be requested on the SASL WG mailing list.
1313 "IETF Consensus" registration procedure MUST be used for registering
1314 new mechanisms in this family. The SASL mailing list
1315 <ietf-sasl@imc.org> (or a successor designated by the responsible
1316 Security AD) MUST be used for soliciting reviews on such
1319 Note to future SCRAM- mechanism designers: each new SCRAM- SASL
1320 mechanism MUST be explicitly registered with IANA and MUST comply
1321 with SCRAM- mechanism naming convention defined in Section 4 of this
1324 IANA is requested to add the following entries to the SASL Mechanism
1325 registry established by [RFC4422]:
1329 Subject: Registration of a new SASL mechanism SCRAM-SHA-1
1331 SASL mechanism name (or prefix for the family): SCRAM-SHA-1
1332 Security considerations: Section 7 of [RFCXXXX]
1333 Published specification (optional, recommended): [RFCXXXX]
1334 Person & email address to contact for further information:
1335 IETF SASL WG <ietf-sasl@imc.org>
1336 Intended usage: COMMON
1337 Owner/Change controller: IESG <iesg@ietf.org>
1343 Menon-Sen, et al. Expires February 1, 2010 [Page 24]
1345 Internet-Draft SCRAM July 2009
1349 Subject: Registration of a new SASL mechanism SCRAM-SHA-1-PLUS
1351 SASL mechanism name (or prefix for the family): SCRAM-SHA-1-PLUS
1352 Security considerations: Section 7 of [RFCXXXX]
1353 Published specification (optional, recommended): [RFCXXXX]
1354 Person & email address to contact for further information:
1355 IETF SASL WG <ietf-sasl@imc.org>
1356 Intended usage: COMMON
1357 Owner/Change controller: IESG <iesg@ietf.org>
1361 This document also requests IANA to assign a GSS-API mechanism OID
1399 Menon-Sen, et al. Expires February 1, 2010 [Page 25]
1401 Internet-Draft SCRAM July 2009
1404 11. Acknowledgements
1406 This document benefited from discussions on the SASL WG mailing list.
1407 The authors would like to specially thank Dave Cridland, Simon
1408 Josefsson and Jeffrey Hutzelman for their contributions to this
1455 Menon-Sen, et al. Expires February 1, 2010 [Page 26]
1457 Internet-Draft SCRAM July 2009
1460 Appendix A. Other Authentication Mechanisms
1462 The DIGEST-MD5 [I-D.ietf-sasl-digest-to-historic] mechanism has
1463 proved to be too complex to implement and test, and thus has poor
1464 interoperability. The security layer is often not implemented, and
1465 almost never used; everyone uses TLS instead. For a more complete
1466 list of problems with DIGEST-MD5 which lead to the creation of SCRAM
1467 see [I-D.ietf-sasl-digest-to-historic].
1469 The CRAM-MD5 SASL mechanism, while widely deployed has also some
1470 problems, in particular it is missing some modern SASL features such
1471 as support for internationalized usernames and passwords, support for
1472 passing of authorization identity, support for channel bindings. It
1473 also doesn't support server authentication. For a more complete list
1474 of problems with CRAM-MD5 see [I-D.ietf-sasl-crammd5-to-historic].
1476 The PLAIN [RFC4616] SASL mechanism allows a malicious server or
1477 eavesdropper to impersonate the authenticating user to any other
1478 server for which the user has the same password. It also sends the
1479 password in the clear over the network, unless TLS is used. Server
1480 authentication is not supported.
1511 Menon-Sen, et al. Expires February 1, 2010 [Page 27]
1513 Internet-Draft SCRAM July 2009
1516 Appendix B. Design Motivations
1518 The following design goals shaped this document. Note that some of
1519 the goals have changed since the initial version of the document.
1521 o The SASL mechanism has all modern SASL features: support for
1522 internationalized usernames and passwords, support for passing of
1523 authorization identity, support for channel bindings.
1525 o The protocol supports mutual authentication.
1527 o The authentication information stored in the authentication
1528 database is not sufficient by itself to impersonate the client.
1530 o The server does not gain the ability to impersonate the client to
1531 other servers (with an exception for server-authorized proxies),
1532 unless such other servers allow SCRAM authentication and use the
1533 same salt and iteration count for the user.
1535 o The mechanism is extensible, but [hopefully] not overengineered in
1538 o Easier to implement than DIGEST-MD5 in both clients and servers.
1567 Menon-Sen, et al. Expires February 1, 2010 [Page 28]
1569 Internet-Draft SCRAM July 2009
1572 Appendix C. Internet-Draft Change History
1574 (RFC Editor: Please delete everything after this point)
1578 o Converted the source for this I-D to XML.
1580 o Added text to make SCRAM compliant with the new GS2 design.
1582 o Added text on channel binding negotiation.
1584 o Added text on channel binding, including a reference to RFC5056.
1586 o Added text on SCRAM as a GSS-API mechanism. This noted as not
1587 relevant to SASL-only implementors -- the normative references for
1588 SCRAM as a GSS-API mechanism are segregated as well.
1592 o Updated References.
1594 o Clarified purpose of the m= attribute.
1596 o Fixed a problem with authentication/authorization identity's ABNF
1597 not allowing for some characters.
1599 o Updated ABNF for nonce to show client-generated and server-
1602 o Only register SCRAM-SHA-1 with IANA and require explicit
1603 registrations of all other SCRAM- mechanisms.
1607 o Removed hash negotiation from SCRAM and turned it into a family of
1610 o Start using "Hash Function Textual Names" IANA registry for SCRAM
1613 o Fixed definition of Hi(str, salt) to be consistent with [RFC2898].
1615 o Clarified extensibility of SCRAM: added m= attribute (for future
1616 mandatory extensions) and specified that all unrecognized
1617 attributes must be ignored.
1623 Menon-Sen, et al. Expires February 1, 2010 [Page 29]
1625 Internet-Draft SCRAM July 2009
1628 o Changed the mandatory to implement hash algorithm to SHA-1 (as per
1631 o Added text about use of SASLPrep for username canonicalization/
1634 o Clarified that authorization identity is canonicalized/verified
1635 according to SASL protocol profile.
1637 o Clarified that iteration count is per-user.
1639 o Clarified how clients select the authentication function.
1641 o Added IANA registration for the new mechanism.
1643 o Added missing normative references (UTF-8, SASLPrep).
1645 o Various editorial changes based on comments from Hallvard B
1646 Furuseth, Nico William and Simon Josefsson.
1650 o Update Base64 and Security Glossary references.
1652 o Add Formal Syntax section.
1654 o Don't bother with "v=".
1656 o Make MD5 mandatory to implement. Suggest i=128.
1660 o Seven years have passed, in which it became clear that DIGEST-MD5
1661 suffered from unacceptably bad interoperability, so SCRAM-MD5 is
1662 now back from the dead.
1664 o Be hash agnostic, so MD5 can be replaced more easily.
1666 o General simplification.
1679 Menon-Sen, et al. Expires February 1, 2010 [Page 30]
1681 Internet-Draft SCRAM July 2009
1686 12.1. Normative References
1688 [RFC2104] Krawczyk, H., Bellare, M., and R. Canetti, "HMAC: Keyed-
1689 Hashing for Message Authentication", RFC 2104,
1692 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
1693 Requirement Levels", BCP 14, RFC 2119, March 1997.
1695 [RFC3174] Eastlake, D. and P. Jones, "US Secure Hash Algorithm 1
1696 (SHA1)", RFC 3174, September 2001.
1698 [RFC3454] Hoffman, P. and M. Blanchet, "Preparation of
1699 Internationalized Strings ("stringprep")", RFC 3454,
1702 [RFC3629] Yergeau, F., "UTF-8, a transformation format of ISO
1703 10646", STD 63, RFC 3629, November 2003.
1705 [RFC4013] Zeilenga, K., "SASLprep: Stringprep Profile for User Names
1706 and Passwords", RFC 4013, February 2005.
1708 [RFC4422] Melnikov, A. and K. Zeilenga, "Simple Authentication and
1709 Security Layer (SASL)", RFC 4422, June 2006.
1711 [RFC4648] Josefsson, S., "The Base16, Base32, and Base64 Data
1712 Encodings", RFC 4648, October 2006.
1714 [RFC5056] Williams, N., "On the Use of Channel Bindings to Secure
1715 Channels", RFC 5056, November 2007.
1717 [RFC5234] Crocker, D. and P. Overell, "Augmented BNF for Syntax
1718 Specifications: ABNF", STD 68, RFC 5234, January 2008.
1720 12.2. Normative References for GSS-API implementors
1723 Josefsson, S. and N. Williams, "Using GSS-API Mechanisms
1724 in SASL: The GS2 Mechanism Family", draft-ietf-sasl-gs2-12
1725 (work in progress), April 2009.
1727 [RFC2743] Linn, J., "Generic Security Service Application Program
1728 Interface Version 2, Update 1", RFC 2743, January 2000.
1730 [RFC3962] Raeburn, K., "Advanced Encryption Standard (AES)
1731 Encryption for Kerberos 5", RFC 3962, February 2005.
1735 Menon-Sen, et al. Expires February 1, 2010 [Page 31]
1737 Internet-Draft SCRAM July 2009
1740 [RFC4086] Eastlake, D., Schiller, J., and S. Crocker, "Randomness
1741 Requirements for Security", BCP 106, RFC 4086, June 2005.
1743 [RFC4121] Zhu, L., Jaganathan, K., and S. Hartman, "The Kerberos
1744 Version 5 Generic Security Service Application Program
1745 Interface (GSS-API) Mechanism: Version 2", RFC 4121,
1748 [RFC4401] Williams, N., "A Pseudo-Random Function (PRF) API
1749 Extension for the Generic Security Service Application
1750 Program Interface (GSS-API)", RFC 4401, February 2006.
1752 [RFC4402] Williams, N., "A Pseudo-Random Function (PRF) for the
1753 Kerberos V Generic Security Service Application Program
1754 Interface (GSS-API) Mechanism", RFC 4402, February 2006.
1757 Zhu, L., "Registration of TLS unique channel binding
1758 (generic)", IANA http://www.iana.org/assignments/
1759 channel-binding-types/tls-unique, July 2008.
1761 12.3. Informative References
1763 [I-D.altman-tls-channel-bindings]
1764 Altman, J., Williams, N., and L. Zhu, "Channel Bindings
1765 for TLS", draft-altman-tls-channel-bindings-05 (work in
1766 progress), June 2009.
1768 [I-D.ietf-sasl-crammd5-to-historic]
1769 Zeilenga, K., "CRAM-MD5 to Historic",
1770 draft-ietf-sasl-crammd5-to-historic-00 (work in progress),
1773 [I-D.ietf-sasl-digest-to-historic]
1774 Melnikov, A., "Moving DIGEST-MD5 to Historic",
1775 draft-ietf-sasl-digest-to-historic-00 (work in progress),
1778 [I-D.melnikov-sasl-scram-ldap]
1779 Melnikov, A., "LDAP schema for storing SCRAM secrets",
1780 draft-melnikov-sasl-scram-ldap-02 (work in progress),
1783 [RFC2898] Kaliski, B., "PKCS #5: Password-Based Cryptography
1784 Specification Version 2.0", RFC 2898, September 2000.
1786 [RFC2945] Wu, T., "The SRP Authentication and Key Exchange System",
1787 RFC 2945, September 2000.
1791 Menon-Sen, et al. Expires February 1, 2010 [Page 32]
1793 Internet-Draft SCRAM July 2009
1796 [RFC4510] Zeilenga, K., "Lightweight Directory Access Protocol
1797 (LDAP): Technical Specification Road Map", RFC 4510,
1800 [RFC4616] Zeilenga, K., "The PLAIN Simple Authentication and
1801 Security Layer (SASL) Mechanism", RFC 4616, August 2006.
1803 [RFC4949] Shirey, R., "Internet Security Glossary, Version 2",
1804 RFC 4949, August 2007.
1806 [RFC5246] Dierks, T. and E. Rescorla, "The Transport Layer Security
1807 (TLS) Protocol Version 1.2", RFC 5246, August 2008.
1809 [tls-server-end-point]
1810 Zhu, L., "Registration of TLS server end-point channel
1811 bindings", IANA http://www.iana.org/assignments/
1812 channel-binding-types/tls-server-end-point, July 2008.
1847 Menon-Sen, et al. Expires February 1, 2010 [Page 33]
1849 Internet-Draft SCRAM July 2009
1855 Oryx Mail Systems GmbH
1863 Email: Alexey.Melnikov@isode.com
1869 West Covina, CA 91790
1872 Email: chris.newman@sun.com
1881 Email: Nicolas.Williams@sun.com
1903 Menon-Sen, et al. Expires February 1, 2010 [Page 34]