Re: iSCSI: Use of SRP (draft -04)

[biran@il.ibm.com]

Steve,

What I meant was that one "AuthMethod" will be negotiated instead of
"InitAuth" and "TargetAuth", so mixed methods are not possible. The
one-way / mutual will be decided in the text messages that carry the
process of the selected method, in a manner specific to that method.

SRP is mutual when using the last server's message by which it proves
the knowledge of the password verifier (assuming the verifiers are kept
confidentially by the server). From RFC2945:
"To finish authentication, they must prove to each other that their keys
are identical...
If the server receives a correct response, it issues its own proof to
the client.  The client will compute the expected response using its
own K to verify the authenticity of the server."

We can have the initiator specifies in the first SRP message (with
the U) "TargetAuth=yes" or "TargetAuth=no" which determines whether
the last server message should be sent, and this will be the manner
specific to SRP (while in KERB5 the initiator set the krb_ap_req mutual
flag).

   Regards,
     Ofer



Ofer Biran
Systems and Software
IBM Research Lab in Haifa
biran@il.ibm.com  972-4-8296253


Steve Senum <ssenum@cisco.com> on 01/03/2001 01:14:31

Please respond to Steve Senum <ssenum@cisco.com>

To:   Ofer Biran/Haifa/IBM@IBMIL, ips@ece.cmu.edu
cc:
Subject:  Re: iSCSI: Use of SRP (draft -04)




Ofer:

If SRP is mutual, then I think the draft should state that with
text similiar to the Kerberos method, and also state how
to handle mixed SRB and Kerberos authentication (or disallow it).

Also, I am not sure I agree that SRP is entirely mutual.
See draft-ietf-pppext-eap-srp-00.txt for a proposal
for using SRP with PPP.

Regards,
Steve Senum

biran@il.ibm.com wrote:
>
> Steve,
>
> You are correct, we'll change the SRP message sequence similar to telnet
(U
> --- N,g,s -- A -- B...).
>
> For simultaneous authentication processes (InitAuth, TargetAuth) it seems
a
> problem of over flexibility. The simpler
> and reasonable way would be to negotiate one authentication method
> AuthMethod and leave the one way / mutual
> authentication decision to the specific method selected. In KERB5 the
> client decides it by setting the krb_ap_req mutual
> flag, in SRP it's actually mutual.
>
>   Regards,
>       Ofer
>
> Ofer Biran
> Systems and Software
> IBM Research Lab in Haifa
> biran@il.ibm.com  972-4-8296253
>
> Steve Senum <ssenum@cisco.com> on 02/28/2001 01:41:01 AM
>
> Please respond to Steve Senum <ssenum@cisco.com>
>
> To:   ietf-ips <ips@ece.cmu.edu>
> cc:
> Subject:  iSCSI: Use of SRP (draft -04)
>
> Julian:
>
> With respect to use of the SRP protocol for authentication,
> I think the current draft is incomplete.  The SRP spec
> requires that values for the Prime Modulus value 'N' and the
> Generator value 'g' be sent by the authenticating entity
> as well as 's' and 'B' (or known through some other method).
> Look at RFC 2944 to see how telnet handles this.
>
> Also, if both Initiator and Target choose to authenticate with
> SRP, or if InitAuth=KERB5 and TargetAuth=srp, the same key names
> will be needed by both sides at the same time, resulting in the
> same key name appearing twice in the same text message.  This
> will make it difficult for the receiver to know which key names
> goes with which authentication process, since there can be two
> going on at one time.
>
> Regards,
> Steve Senum