Re: [Supa] SUPA Update

["Joel M. Halpern" <[e-mail]>]

I have not reviewed the Chen draft until now.
A brief review shows that it does not follow the patterns or approach 
defined in the IETF SUPA Information Model and the correspnding generic 
data model.

This matters as those models use exactly the strong separation you ask for.

Yours,
Joel

On 7/5/17 10:39 PM, youlizhao wrote:
> Dear All,
> 
> Below are my thought on ECA DM based on the materials I read and the 
> project I am doing. Any comments would be welcome.
> 
> The discussions are based on an ECA Policy DM draft [eca-data-model]. 
> According to our understanding, it is aligned with the ECA Policy IM 
> [eca-info-model].
> 
> We appreciate that the draft [draft-chen-supa-eca-data-model-05] 
> provides the entity/script abstraction that allows extension for 
> different scenarios, and they match our implementation. In particular, 
> our ECA policy parser can read entity fields (or script fields), and map 
> them to the CLI command parameters (or script implementation).
> 
> However, we found that the current definition still makes DM parsing 
> complicate. Note that, in [draft-chen-supa-eca-data-model-05], the 
> condition-list clause may still involve with network details (e.g., 
> bandwidth and threshold in the service-flow example), and the 
> action-list clause may still contain the decision logic (i.e., when the 
> bandwidth exceeds a threshold) to trigger an action. To extract the 
> “real” condition, our policy parser needs to understand the action-list 
> clause, and then generate the “condition” logic in real implementation.
> 
> To make things more complicate, sometimes we need to judge several 
> conditions to trigger an action (e.g., in Cisco EEM [CiscoEEM, Pages 69, 
> 342, 368] and some SD-WAN examples in Huawei SD-WAN solution 
> [Huawei-SD-WAN]). In addition, some constraints may be added, e.g., the 
> conjunctive-type duration that indicates how long the conjunctive 
> relationship should hold. However, the current model is weak on 
> expressing the capability (although conjunctive-type is defined in 
> [eca-data-model], how to use it is not clear).
> 
> Based on the above arguments, we would like to suggest the following 
> changes:
> 
> (a) A clear separation of events, conditions, and actions is welcome so 
> that our parser can just care different parts and directly map them to 
> implementation;
> 
> (b) The condition clause can be enriched to express the 
> conjunctive-type, conjunctive-duration, and other possible fields.
> 
> Thanks!
> 
> Regards,
> 
> Leo
> 
> [eca-info-model] Generic Policy Information Model for SUPA
> 
> https://tools.ietf.org/html/draft-strassner-supa-generic-policy-info-model-05 
> 
> 
> [eca-data-model] ECA Policy YANG Data Model
> 
> https://tools.ietf.org/html/draft-chen-supa-eca-data-model-05
> 
> [Cisco-EEM] Embedded Event Manager Configuration Guide,
> 
> http://www.cisco.com/c/en/us/td/docs/ios-xml/ios/eem/configuration/12-2sx/eem-12-2sx-book.pdf
> 
> [Huawei-SD-WAN] Huawei SD-WAN Solution Brochure
> 
> http://e.huawei.com/en/solutions/technical/sdn/enterprise-wan/sd-wan
> 
> *From:*SUPA [mailto:[e-mail]] *On Behalf Of *youlizhao
> *Sent:* 2017年6月20日15:38
> *To:* Benoit Claise <[e-mail]>; King, Daniel 
> <[e-mail]>
> *Cc:* [e-mail]; [e-mail]; SUPA list <[e-mail]>
> *Subject:* Re: [Supa] SUPA Update
> 
> Dear Benoit AD,
> 
> Thanks for pushing and monitoring the SUPA WG progress.
> 
> I am sorry to see the email below. I am an engineer in Huawei, and 
> recently we have been developing the code of SUPA ECA policy to control 
> the network. In particular, we provide restful interface to users for 
> configuring ECA policies following the ECA DM.
> 
> To be honest, IMO, the ECA IM and DM are not mature enough for direct 
> implementation. We are revising some parts facilitating the implementation.
> 
> Although the SUPA IM and DM still contain flaws, we hoped to see them 
> completed (also other documents, e.g., SUPA framework) so that we can 
> complete our own implementations based on SUPA standard work.
> 
> Thanks for your consideration.
> 
> Regards,
> 
> Leo
> 
> *From:*SUPA [mailto:[e-mail]] *On Behalf Of *Benoit Claise
> *Sent:* 2017年6月15日20:47
> *To:* King, Daniel <[e-mail] 
> <mailto:[e-mail]>>; SUPA list <[e-mail] 
> <mailto:[e-mail]>>
> *Cc:* [e-mail] <mailto:[e-mail]>; [e-mail] 
> <mailto:[e-mail]>
> *Subject:* Re: [Supa] SUPA Update
> 
> Dear all,
> 
> After the last IETF, I put a calendar reminder on June 16th to decide on 
> the next steps for SUPA.
> This is inline with the our previous meeting minutes, so it should not 
> come as a surprise.
> Granted, this is one day earlier than foreseen, but the IESG agenda 
> coordination call takes place today, and it was important from a 
> scheduling point of view to understand if SUPA would meet. The chairs 
> informed me that no SUPA meeting is required in Prague. That triggered 
> this discussion, just one day earlier.
> 
> Our meeting minutes: 
> https://www.ietf.org/mail-archive/web/supa/current/msg01612.html
> 
>     At the SUPA WG at IETF 98 (Tuesday, 28 March) we discussed the
> 
>     progress of the WG.  Benoit (our AD) summed up the situation, pointing
> 
>     out that our drafts are not updated very frequently, and that the SUPA
> 
>     mailing list has been very quiet between meetings.
> 
>     At the meeting the authors of the SUPA Information model and the SUPA
> 
>     Data Model drafts said that those drafts should be ready for WG Last
> 
>     Call by 1 June, so that they could be sent to IESG for approval by
> 
>     about 1 July.
> 
>     After summing up the pros and cons for SUPA continuing, Benoit
> 
>     concluded by saying that the WG will be closed at IETF 99 (Prague, 16
> 
>     July) unless there is substantive progress on the Information Model
> 
>     and especially on the Data Model drafts by one month before the Prague
> 
>     meeting.  'Substantive progress' here means seeing comments on and/or
> 
>     reviews of these drafts demonstrating that people - outside the small
> 
>     group of authors - have carefully read the drafts, or better, that they
> 
>     are actually using SUPA's Information and Data Models.
> 
> I've been watching the list.
> 
> Since the last IETF meeting, we received two new drafts ...
> 
>          draft-ietf-supa-generic-policy-info-model-03.txt
>          draft-ietf-supa-generic-policy-data-model-03.txt
> 
> ... and some draft reviews:
> 
>         [e-mail] <mailto:[e-mail]> on on
>         draft-ietf-supa-policy-based-management-framework:
> 
>             Good feedback but it seems like only editorial to me.
> 
>         Tony [e-mail] <mailto:[e-mail]> on
>         draft-cheng-supa-applicability:
> 
>             Some editorial comments and three technical ones:
> 
>             1.       I wonder the meaning of section 3, the part copied
>             from framework draft, may not be needed.
> 
>             2.       I suggest to replace the title of 4.2.2.and 4.2.3
>             with detailed information instead of writing just   Example
>             1 / 2.
> 
>             3.       The writer wrote “We will define "edgeInterface"
>             role and "EnterpriseDomain" later in  this note” but I
>             failed to find the explanation for these two term.
> 
>             Benoit => it's more like one technical comment, the last one.
> 
> 
>         Haining Wang: [e-mail] <mailto:[e-mail]> on
>         draft-ietf-supa-generic-policy-data-model-03:
> 
>             I understand that the GPIM YANG model provides an example of
>             how to convert IM to DM (for general policy), and John’s
>             SNMP blocking example
>             (https://mailarchive.ietf.org/arch/msg/supa/DWEzaSBK6KBdsmQ0FE2-eypTzeY)
>             exposes some details. But I am sorry that the whole picture
>             is still not clear to me. It would be nice if the ECA Data
>             Model part can explain in more details.
> 
>         March Blanchet on draft-ietf-supa-policy-based-management-framework:
> 
>             - larger comment: I’m not sure what to do with this
>             document. It looks like a large wish list of features. I
>             guess I’m probably too used to implementation/protocol
>             details. I guess I will wait until to see the actual
>             protocol/yang models.
> 
> 
> Let's analyze the situation:
> I don't consider those reviews (btw a single one the DM, none on the IM) 
> as "substantive progress".
> I don't see interest from YANG module authors, ready to reuse the SUPA 
> YANG constructs.
> Being a year late according to the charter milestones, the window of 
> opportunity to produce reusable work has been closing rapidly.
> I believe that SUPA had multiple chances to make it happen, and failed 
> to deliver.
> With this in mind, I don't see how I should conclude anything else than 
> this WG will be closing at IETF 99.
> 
> Regards, Benoit (OPS AD)
> 
>     Dear supa’rs,
> 
>     We have cancelled our formal meeting in Prague. This decision was
>     taken based on a proposed plan to focus effort on completing the
>     existing WG items and prepare for closure of the supa working group
>     sometime between IETF 99 and 100. A plan that is yet to be approved
>     by Benoit.
> 
>     During the last working group meeting Benoit stated:
> 
>     “the WG will be closed at IETF 99 (Prague, 16 July) unless there is
>     substantive progress on the Information Model and especially on the
>     Data Model draft by one month before the Prague meeting.”
> 
>     The authors of the Data Model and Information Model I-Ds did submit
>     new versions but we only received one review. However, Nevil and I
>     are working with the IM and DM authors to gather reviewers in
>     preparation of Last Call. Essentially, we are working to prep folks
>     who would be able to review the documents we Last Call, ideally
>     these should be from policy/yang implementers.
> 
>     The Framework I-D has also received a review which is positive, and
>     I am in the process of reviewing the document myself to also help
>     prepare the document for Last Call. Additionally, the Applicability
>     I-D (a non-working group document) received a review which is also
>     useful.
> 
>     We have also seen notifications from other SDOs following supa,
>     specifically:
> 
>     - ONUG: Investigating I2NSF combined with the SUPA data model and
>     framework
> 
>     - ETSI Experiential Networked Intelligence (ENI): New initiative
>     defining context aware networking systems, SUPA was identified as a
>     key building block
> 
>     - MEF Open Lifecycle Service Orchestrator (LSO): Using SUPA between
>     functional components
> 
>     However, the indication from ONUG, ETSI and MEF does not materially
>     change the situation of SUPA but it does demonstrate wider interest
>     in our work, and at least some responsibility for supa/IETF to
>     complete it (if possible). If you are aware of near-term
>     implementations now is the time to highlight them.
> 
>     Again, we felt we did not need a WG meeting in Prague to progress
>     the working group I-Ds, and given the IETF agenda coordination call
>     (is today) we had to cancel the supa WG session request ASAP, and
>     unfortunately before we had a chance to communicate the current
>     situation to the rest of the working group. Apologies for any
>     surprise when you saw the cancellation notification, and the lack of
>     opportunity for wider discussion.
> 
>     As mentioned our proposed plan has been submitted to Benoit and is
>     yet to be approved, therefore we will wait for his thoughts and
>     ultimate decision.
> 
>     The SUPA Chairs would sincerely like to thank everyone for their
>     participation and especially the authors of I-Ds for their efforts.
> 
>     BR, Nevil and Dan.
> 
> 
> 
> _______________________________________________
> SUPA mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/supa
>