|
Hi Violeta, I am fine with the latest version of the document. Thank you, Yaron On 06/14/2011 09:41 PM, Cakulev, Violeta (Violeta) wrote: Hi Yaron, Thanks for the suggestions. Please see inline. -Violeta -----Original Message----- From: Yaron Sheffer [mailto:yaronf.ietf@xxxxxxxxx] Sent: Friday, June 03, 2011 4:32 PM To: Cakulev, Violeta (Violeta) Cc: ietf@xxxxxxxx; IPsecme WG; dime@xxxxxxxx; draft-ietf-dime-ikev2-psk-diameter@xxxxxxxxxxxxxx Subject: Re: [IPsec] Last Call: <draft-ietf-dime-ikev2-psk-diameter-06.txt> (Diameter IKEv2 PSK: Pre-Shared Secret-based Support for IKEv2 Server to Diameter Server Interaction) to Proposed Standard Hi Violeta, thanks for your response. I understand now where you're coming from. But the next person to read the document might not have the authors so readily available :-) Implementors need to understand the motivation for this solution as well as the missing pieces. [VC] And the next person indeed had the same concern. Therefore, in v8 we made the changes as proposed below. So I would suggest that you add: - A short paragraph in the Introduction putting this document in perspective and referencing (non-normative references of course) the 3GGP2 documents. - Another paragraph in the Security Considerations that says that generation/derivation of the PSK is security-critical, and provides the 3GPP2 docs again as an example of a solution to this problem. [VC] Please see v8. We added statements both in Introduction and Security Considerations as proposed above. Regarding usage of the SPI, the document says: "Upon receiving the IKE_AUTH message from the IKEv2 Peer, the IKEv2 Server uses the information received in IDi *and the SPI* if available, to determine if it has the PSK for this IKEv2 Peer."This sounds to me as if the SPI is used as part of the PSK lookup operation. Maybe the SPI should not be mentioned in the above text. [VC] We modified this paragraph in v8. Please see v8. Best regards, Yaron On 06/01/2011 11:04 PM, Cakulev, Violeta (Violeta) wrote:Hi Yaron, Thanks for the comments. Please see inline [VC]. -Violeta -----Original Message----- From: ipsec-bounces@xxxxxxxx [mailto:ipsec-bounces@xxxxxxxx] On Behalf Of Yaron Sheffer Sent: Sunday, May 22, 2011 2:38 PM To: ietf@xxxxxxxx Cc: IPsecme WG; draft-ietf-dime-ikev2-psk-diameter@xxxxxxxxxxxxx; dime@xxxxxxxx Subject: Re: [IPsec] Last Call:<draft-ietf-dime-ikev2-psk-diameter-06.txt> (Diameter IKEv2 PSK: Pre-Shared Secret-based Support for IKEv2 Server to Diameter Server Interaction) to Proposed Standard Hi, Having read this document only now, I think there's a number of serious issues with it. This document was sent to the ipsec mailing list a while ago but unfortunately got no review. Summary: 1. I think the wrong architectural choice was made, in preferring PSK over EAP authentication. 2. There is not enough detail in the document to result in interoperable implementations. [VC] Please see responses to detailed comments. Detailed comments: * The appropriate ref for IKEv2 is RFC 5996. This was actually noted in the shepherd review back in March. [VC] This is fixed in v7. * The document notes that EAP is one of the authentication modes supported by IKEv2. EAP is designed for interaction with backend AAA servers, and is quite capable of performing shared-secret authentication, using a variety of EAP methods (and see also RFC 5998, on IKEv2 mutual auth with EAP). Yet the document does not explain why EAP is not used, instead preferring the IKE PSK authentication method. [VC] Diameter application for Diameter client to server communication for IKEv2 with EAP has been specified in RFC 5778. However, there is no Diameter application specified for IKE PSK authentication method. This is stated in the draft both in Abstract and Introduction. The draft does not recommend one or the other, it is simply specifying what is not specified. It is up to a specific use case to decide what to use. For example 3GPP2 decided to use IKEv2 PSK (in 2 of its specifications), and that is what triggered this effort in IETF. * 4.1: how can the incoming SPI be used to identify the peer? [VC] As stated in Section 4.1 IDi payload is extracted from the IKE_AUTH message and this payload 'MUST contain an identity that is meaningful for the Diameter infrastructure, such as a Network Access Identifier (NAI), since it is used by the IKEv2 Server to populate the User-Name AVP in the Diameter message'. SPI is not used to identify the peer. * Packing additional semantics into SPI may conflict with elements of the IPsec architecture (see for example Sec. 9.3 of draft-ietf-ipsecme-failure-detection-08). [VC] Agreed. * 4.1, 2nd paragraph: generation of the PSK is central to this solution, so it cannot be "outside the scope" of the document. There is no way to interoperate otherwise. [VC] This draft focuses on IKEv2 server, as a Diameter client, to the Diameter server communication for IKEv2 with pre- shared secret based authentication, and specifies Diameter application for this communication. It is left to specifications that make use of this Diameter application to specify where the PSK comes from and how it is generated. As mentioned above, there are two 3GPP2 specs that make use of this Diameter application and both of those specs specify generation of the PSK. * Moreover, if a single client is expected to sometimes use EAP and sometimes PSK, there must be a way to notify it which one to use. [VC] This is not the assumption. Again this draft specifies Diameter application for Diameter client to the Diameter server communication for IKEv2 with PSK. IKEv2 server knows whether EAP or PSK is used. * How does key-lifetime relate to the lifetime of the IKE SA? [VC] This should be the same as in RFC 5996, how pre-shared secret lifetime relates to the lifetime of the IKE SA. This draft should not make any changes. * Sec. 10 refers to the PSK as a "session key" which is incorrect, as PSK is only used for authentication and does not encrypt anything. [VC] Good point. It is changed to PSK in v7. * The same paragraph is very vague about the security properties of PSK. RFC 5996 takes PSK much more seriously, e.g. "When using pre-shared keys, a critical consideration is how to assure the randomness of these secrets." Again, I believe the document should specify how the PSK is derived. [VC] I absolutely agree that derivation of PSK is critical. However, as stated above, this draft specifies Diameter application only. Therefore, security consideration section cannot include much more details on derivation of PSK as it is specified elsewhere. * Why "if nonces are included" where the document says that they *must* be included (in the AVP occurrence table). [VC] Good point. This is removed in v7. Thanks, Yaron On 05/20/2011 04:50 PM, The IESG wrote: |
_______________________________________________ Ietf mailing list Ietf@xxxxxxxx https://www.ietf.org/mailman/listinfo/ietf