Re: [secdir] (Security sections) SecDir and AppsDir review of draft-ietf-storm-iscsi-cons-06

[Alexey Melnikov <[e-mail]>]

Hi David,

On 09/10/2012 17:14, Black, David wrote:
>> Hi Alexey, here are the responses to your comments specific to security
>> sections of iSCSI consolidated draft - actually, I am deferring mostly to
>> Julian and David who are better suited than me to comment on this area, :-)
> That would be my cue ... inline ...
>
>>> In 9.3.1:
>>>
>>> - HMAC-SHA1 MUST be implemented [RFC2404].
>>>
>>> RFC 2404 seems to define HMAC-SHA-1-96, not HMAC-SHA1.
>> [Mallikarjun:] That is true. I do not know the reason for this citing.
>> Julian/David?
>>
>> I also found it interesting that the abstract for 2404 itself does not use the
>> "96" qualifier.
> IPsec uses HMAC-SHA1 with its output truncated to 96 bits.  HMAC-SHA1 was
> used here as being a more recognizable algorithm name, but the specific
> requirements of RFC 2404 do apply.  Here's some revised text that handles
> both concerns:
>
>   - HMAC-SHA1 MUST be implemented in the specific form of HMAC-SHA-1-96 [RFC2404].

I like this. Thanks.

>>> 9.3.2. Confidentiality
>>>
>>>     The NULL encryption algorithm MUST also be implemented.
>>>
>>> I find it odd that the section talks about how weak DES is and then
>>> requires NULL encryption to be supported. What is the reason for this?
>>   [Mallikarjun:] IIRC, I *think* this was because we wanted implementations to
>> be able to use the authentication/MAC of IPSec suite, without forcing them
>> always to use encryption. David, can you please add/correct?
> Mallikarjun is basically correct, but there's more to explain.
>
> The NULL encryption algorithm is needed to allow use of ESP for authentication
> (cryptographic integrity) without encryption.  This is often preferred to AH
> for that purpose, especially in hardware implementations.

Ok.

>>> 9.3.3. Policy, Security Associations, and Cryptographic Key
>>>           Management
>>>
>>>        - When digital signatures are used to achieve authentication,
>>>          an IKE negotiator SHOULD use IKE Certificate Request
>>>          Payload(s) to specify the certificate authority. IKE
>>>          negotiators SHOULD check the pertinent Certificate
>>>          Revocation List (CRL) before accepting a PKI certificate for
>>>          use in IKE authentication procedures.
>>>
>>> What are the reasons for these requirements being SHOULD level (as
>>> opposed to MUST level)?
> There are environments in which a small number of certificates are statically
> configured as trust anchors in which these mechanisms may not be needed.

I think mentioning this would be useful.

>>>     - The following identification type requirements apply to IKEv1.
>>>       ID_IPV4_ADDR, ID_IPV6_ADDR (if the protocol stack supports
>>>       IPv6) and ID_FQDN Identification Types MUST be supported;
>>>       ID_USER_FQDN SHOULD be supported. The IP Subnet, IP Address
>>>       Range, ID_DER_ASN1_DN, and ID_DER_ASN1_GN Identification Types
>>>       SHOULD NOT be used. The ID_KEY_ID Identification Type MUST NOT
>>>       be used.
>>>
>>> It would be good to know the reason for the last SHOULD NOT and the last
>>> MUST NOT.
>>   [Mallikarjun:] I will defer to Julian and David on these.
> Sure ... this was done back in RFC 3270 and is being carried forward
> from there (i.e., none of these requirements are new).
>
> IP Subnet and IP Address Range are too broad to usefully identify an
> iSCSI endpoint, hence they are "SHOULD NOT".
>
> The _DN and _GN types are X.500 identities; unless one is a PKI
> expert, the better approach is usually to use subjectAltName.
> The primary reason for the "SHOULD NOT" was to warn those who
> are not PKI experts away from X.500 identities.
>
> ID_KEY_ID is not interoperable as specified in RFC 2407
> ("opaque byte stream which may be used to pass vendor-specific
> information"), hence they are "MUST NOT".
>
> Should explanatory text for these be added to the draft?

I think this would be useful. It might also help you with revising the 
document in the future.