Minutes from Jean Mahoney. Thank you very much! IETF-99 DIME 1740-1840 July 17, 2017, Monday Afternoon session III meeting room: Karlin III Jabber room: dime at jabber.ietf.org (Please join) Chairs:Jouni Korhonen, Lionel Morand AD: Ben Campbell 17:40 - 17:45, Preliminaries, Chairs (5 minutes) ------------------------------------------ Presentation: https://www.ietf.org/proceedings/99/slides/slides-99-dime-agenda-wg-status-00.pdf slide 1: Title Lionel Morand: Jouni is not able to make it to the meeting. slide 2: Note Well slide 3: Agenda Note taker/Jabber scribe: Jean Mahoney Martin Dolly: when will Diameter security discussion happen? Lionel: end of the session. 17:45 - 17:50, WG Document Status, Chairs (5 minutes) ------------------------------------------ *draft-ietf-dime-agent-overload-11 --> RFC Ed Queue *draft-ietf-dime-doic-rate-control-06 --> Waiting for Write-Up *draft-ietf-dime-group-signaling-08 --> In WG Last Call *draft-ietf-dime-load-09 --> RFC Ed Queue *draft-ietf-dime-rfc4006bis-03 --> In WG Last Call slide 4: WG Status Update Lionel: Both overload and load are in the RFC editor queue due to a MISREF. I'll take care of it after this meeting. Steve Donovan: Just a nit on the slide - I'm the editor on load information, I'm not the editor for 4006bis. Lionel: Group signaling - the author's think it's ready. We need more reviews. Marco? Marco Liebsch: we've updated it before Chicago. We've covered all comments. Is ready for expert review. Lionel: we've received a good technical review from Steve. Needs another before WGLC. Rate control is missing its writeup. Steve: Has it finished WGLC? Lionel: Yes, needs writeup. Rate-control didn't receive a deep technical review. Need further review. From SA5 guys in 3GPP, since they are the main users. slide 5: WG Status update (2/2) Lionel: AVP level security (e2e) draft has expired - Jouni is not able to support the workload alone. I will go to the SAAG meeting and will ask for support from security experts. Solution doesn't have to be Diameter specific. Ben Campbell: I would personally like to see this completed. Who will deploy it? Martin Dolly: next round of FCC sys req, they just finished looking at SS7 security vulnerabilities, next 2 years they'll look at Diameter. SA3 coming up in 3 weeks. I'll give them discussion paper, a status update of work here. It gets friendly talk there. It's been discussed with regard to 5G and security system. It's now or never, forever-hold-your-peace time. Lionel: GSMA would like a recommendation. If it's easy to implement, we have ongoing deployment. Ben: They are welcome to come here and help. Eric Guttman: for mission-critical deployments, there are 2 security domains - operator domain and mission-critical operator domain. We've looked at securing at a message or component to a SIP message level. The message has to be e2e secure between client and host at the mission-critical ISP. Are there AAA requirements where you can't expose certain AVPs to the operator? Need to check. Lionel: From the requirements POV, both should be available with the solution. Normally you would only be able to secure signaling between gateways for instance. You may have e2e secure exchange between any Diameter client and any Diameter server. Martin: I don't think that's a case. 401 specifies that equipment manufacturers must support IPsec for communication within the EPC, but says it optional to deploy. No one has deployed. For 5G, the European operators are fighting pulling the security into the EPC not at the ENodeB. Some of those statements aren't true. Lionel: Ben, you see there is some interest. We would like to see this work done in IETF. Ben: I observe an interest in using the work. I need to observe interest in doing the work. Maybe SAAG will help. Lionel: It's more a security discussion than a Diameter discussion. Need help from security area. Eric: There was a mechanism for using an SMIME encapsulation. That became historic? Lionel: deprecated. A new mechanism will be defined outside the base protocol. All new drafts are based on this. 17:50 - 18:35 Individual draft discussions (45 minutes) ------------------------------------------- 17:50 - 18:10 Diameter Policy Groups and Sets, Lyle (20min) Draft: https://tools.ietf.org/html/draft-bertz-dime-predictunits-02 Draft: https://tools.ietf.org/html/draft-bertz-dime-policygroups-04 Presentation: https://www.ietf.org/proceedings/99/slides/slides-99-dime-predicted-units-policy-groups-00.pdf slide 1: Title slide 2: Predicted Units v02 - Motivation Lyle: It takes 3-4 minutes to spin up a VM. slide 3: Predicted Units v02 - Update Lyle: Have not received any new updates. I would like to ask for adoption. Lionel: Who has read? I wasn't in the last meeting what was the feeling in the room? Ben: Besides the chair and the AD? Jean: About 7 people total. Lionel: I've read it and understood the mechanism, which is good. Check on the ML if there is support for adoption. There's not much to do on this doc. slide 4: Intermission slide 5: Purpose slide 6: Policy Groups - Update slide 7: Relationship Model slide 8: Policy Groups Example 1, Overlap Deduplication at Enforcement Point - Adding Membership Assignment to Filters Lyle: SDN switches have limitations of tables (13 tables in OpenFlow). Can concatenate tables. The rules no longer overlap. How to use your metadata field and tie it back to a customer? slide 9: Policy Groups Example 2, Application at the Decision Point Process Lyle: We don't change the user's relationship to their services mid-session. Usually reauthorize them before doing that. slide 10: Applications slide 11: Next Steps Lyle: Would like to hear more from the group before asking for adoption. Lionel: who has read? Lionel: Go back to the Purpose slide. Lyle: didn't want to confuse people but wanted a similar aggregation concept with backwards compatibility. It's similar to ... It's a membership domain. Lionel: I need to review it again. It would be useful to have more review. Lyle: I'm not happy with the examples, I'd like them to be consistent through the doc. Lionel: the draft needs to clearly identify why you can't do this with existing mechanisms. Lyle: all of our filters are based on time and what's in the packets. When the time is the same and the rule is a default rule, that's the most overlap in the system. Lionel: It's not that clear from the creation of AVP, be good to describe it. 18:10 - 18:35 Diameter Specification Recommendations, Lyle (25min) Draft: https://tools.ietf.org/html/draft-bertz-dime-diamimpr-00 Pres: https://www.ietf.org/proceedings/99/slides/slides-99-dime-diameter-specification-recommendations-00.pdf slide 1: Title slide 2: Motivation Lyle: 85% of the time, errors due to ambiguity. slide 3: Findings Eric Guttman: send this to SA5 charging if you have a delegate. Lyle: we don't, but wanted to get it out. slide 4: Methodology slide 5: Unexpected Use Cases slide 6: We are not immune slide 7: Enumerations slide 8: Recommendations slide 9: Defined AVP Recommendations slide 10: Defined AVPs Recommended Formats slide 11: Imported AVP Recommendations slide 12: Grouped AVP / Command Refinement slide 13: Example Refinement slide 14: Command Recommendations slide 15: Enumeration Recommendations slide 16: GAPs for Automated Validation slide 17: Example to Close Gaps slide 18: Enumeration Example Format slide 19: Open Questions slide 20: Summary slide 21: Next steps Lionel: who has read? (Three people raised hands) Lyle: Dave Dolson and Jouni read it. Lionel: in 3GPP it's not really refinement, because in 3GPP just highlight the AVPs they use. We you take a specification and not reading anything - Lyle: they want to just write the code and be done. Lionel: For new drafts, it should be taken into consideration. For existing RFCs, these are documentation errors. When it's clarification, it needs to go through the errata process. For the 3GPP specs, CRs need to be opened. Lyle: This has a lot of errors documented. Would this be informational? How would it progress? Lionel: It can be informational, and can be referenced by other docs. No guideline here. Just highlight what could go wrong. Lyle: I would change the style. Make it more of a report. Lionel: It's not unusual for the IETF to explain issues with implementations. 18:35 - 18:40 Wrap-up and Next, Chairs/AD (5 minutes) ----------------------------- WG Goals/Milestones status - skipped Next steps/Action Points - Future of the WG Lionel: We have 3 remaining documents. Ben: And two that you are thinking of adopting. Lionel: few people in room, not much activity on the list. I won't be able to continue with the chair position - too much work in the 3GPP. Same with Jouni - he changed companies, won't be able to continue. What is the next step except closing the group? Ben: The existing documents are post or near WGLC. Those can be in the hands of the ADs before long. What's you're time frame for needing to step down? Lionel: I need to know when to stop adopting new documents. Even for Jouni. Ben: Of the 2 drafts for the adoption - the predictive units can be AD sponsored. The group one is on the border, a little more complex. I don't have to be the AD that sponsors it. Lionel: For the group one, I need to see more interest. Ben: even if I AD sponsor, I want someone to review them. Lionel: E2E security needs to be done. If we get volunteers from SAAG. Ben: lets see what happens in SAAG. If there's interest - maybe we can spin up some mini-group. Or just speculating here - recharter dime with new chairs to just do that. Lionel: If it is only to go with existing drafts, I think it can be done in short timeframe. Need to confirm with Jouni. Defining new charter with new chairs may not be useful. Ben: E2E security may go into the security area. Lionel: Are there any other comments? (none from the room). To summarize - existing WG documents will be pushed forward. For predictive units, confirm on ML that it's a WG doc, for the group one, we'll check. After those documents, we'll close the WG. Ben: Those 2 new drafts - predictive definitely and maybe group - can be AD sponsored. They can still change to AD sponsorship if we need to close the working group sooner. I'm skeptical of bringing new work into the group, unless we see interest. Lionel: Thank you. ACTION: Chairs to push existing WG documents forward. ACTION: Lionel to bring the e2e security draft to the SAAG session this week to ask for support in finishing it. ACTION: Confirm interest on mailing list for adopting draft-bertz-dime-predictunits. ACTION: Confirm interest on mailing list for adopting draft-bertz-dime-policygroups. ACTION: Ben to look more closely at draft-bertz-dime-policygroups as a candidate for AD sponsorship.