[Ibnemo] 答复: [Supa] 答复: SUPA call: 7:30-8:30am, Monday, July 13, San Francisco time

[Xiayinben <[e-mail]>]

Hi John,

Thanks for your question  for nemo's object model.

Yes, most of today's network model and properties/configuration are very complex,  is this not the problem what we want to solve?
let's think about why we have so complex configuration today. I can summary these configuration to two groups. One group is internal configuration to implement a network level function, e.g. configure PE,P to build a tunnel. Second group is for external features, e.g. config TE bandwidth. This group is really what customers care. Today, we have SDN and we can get network level function rather than device configuration one by one. So can we forget first group configuration? Do we need to define new network level objects and properties?

VPN is a specific network level service. SFC is a specific network level service. We can provide specific NBI for each specific network level service. But in future, we will have more and more network level service, such as new VN which can support any topology (VPN only supports fullmesh or hub-sopken topology). These network level service NBI will grow forever.
so I think Intent NBI should define network level mate objects rather than network level services. By using these mate objects, we can composite network level services. Neither VPN nor VN is combination of nodes and connections, so I think user can use node, connection to manage a simple VPN requirement. SFC is combination of nodes, connections and flows. So we summarize that network level mate objects are  three simple objects -node, connection and flow.

The properties of node, connection, flow may be lots. But those are cared and can be understood by users. What we discussed in ibnemo is to confirm those objects and its properties as much as possible. If we can cover 80% scenarios requirements, I think that will be useful.

Looking forward your comments.

Yinben


发件人: John Strassner
发送时间: 2015年7月20日 17:23
收件人: Xiayinben; John Strassner
抄送: [e-mail]; [e-mail]; Susan Hares; Scott O. Bradner
主题: RE: [Supa] 答复: SUPA call: 7:30-8:30am, Monday, July 13, San Francisco time

Hi Yinben,

The reason I brought up the complexity in previous network managed object specifications is that I don't understand how your three simple abstractions (node, link, and flow) can be used to manage even simple concepts, like a VPN. Look at the complexity in CTPs and PTPs for example. Look how G.805 and G.809 conflict with each other. I simply don't see how you can hide all of the semantics with these concepts by just saying "I have a link" or "I have a node". What I see happening is that people will subclass your three basic classes (which will ruin interoperability), not use them because they can't find what they want (which ruins the effort), or you will be forced to add a LOT of properties to these three base objects. This last approach violates good OO design practice. Instead, I recommend that you have a more robust set of classes and use metadata to augment your base classes with semantics. There are many different OO patterns that you could use for this purpose.

You are correct - GPIM is ONLY about abstracting policy. We specifically avoided defining models for generic resources and services - those are not in the SUPA charter, and we would much rather use the work of other WGs. That's why I continue to bug you about your network model. :-)

Finally, work has started on the SUPA data models. Stay tuned.


regards,
John

John Strassner, Ph.D.
CTO, Software Laboratory, CRD
[logo.gif]


Futurewei Technologies
US R&D Center
2330 Central Expressway
Building A, office A2-2143
Santa Clara, California   95050

  Office:  +1.408.330.4923
  Email:   [e-mail]<mailto:[e-mail]>





From: Ibnemo [mailto:[e-mail]] On Behalf Of Xiayinben
Sent: Sunday, July 19, 2015 7:09 PM
To: John Strassner
Cc: [e-mail]<mailto:[e-mail]>; [e-mail]<mailto:[e-mail]>; Susan Hares; Scott O. Bradner
Subject: [Ibnemo] 答复: [Supa] 答复: SUPA call: 7:30-8:30am, Monday, July 13, San Francisco time

Hi John,

Ibnemo’s target is to simplify network interface. I think this is also the interesting of industry for Intent NBI.
Intent NBI is based on some level of network abstraction. An appropriate abstraction provides a simple and intuitive network interface and hides implementation detail at same time.
The network object model and network operation model in Ibnemo represent our attempt. They certainly need polish. That is why we need Ibnemo.

I know you love policy so much. I can understand you want policy theory to be more perfect and can cover more range, just as your effort in GPIM https://tools.ietf.org/html/draft-strassner-supa-generic-policy-info-model-02.
I think GPIM is a real generic model. Actually, it is not limited in network area.  I didn’t find any specific network abstraction in it.
I wish GPIM would have tremendous progress than RFC3460. But I think we still need concrete network abstraction for Intent NBI.

Yinben


发件人: John Strassner [mailto:[e-mail]]
发送时间: 2015年7月17日 7:47
收件人: Xiayinben
抄送: Tina TSOU; Scott O. Bradner; Susan Hares; [e-mail]<mailto:[e-mail]>; [e-mail]<mailto:[e-mail]>
主题: Re: [Supa] 答复: SUPA call: 7:30-8:30am, Monday, July 13, San Francisco time

What, exactly, are you not agreeing with in the following sentence:
"NEMO uses a flatter, simpler object model with fewer objects to represent targets of policy. NEMO does not define a policy model, and does not support ECA policies. NEMO uses a condition-action paradigm to execute its policy commands.”

Lets take the sentence point by point.

#1 FLATTER, SIMPLER OBJECT MODEL
NEMO defines three objects: node, link, and flow. How many managed objects do G.805 or G.809, let alone the average MIB, define? It is irrelevant that you can define subclasses of node - the user has to do all the work, and this guarantees non-interoperable solutions.

#2 TARGETS OF POLICY.
You just said that you define an object model and an operation model. Both would be policy targets, by any standard CA or ECA definition.

#3. NEMO DOES NOT DEFINE A POLICY MODEL
You said: "NEMO doesn’t NEED to define a new policy model, because industry already has well defined policy model(such as RFC 3460). We will not rebuild a wheel." Not only did you agree that you aren't defining a policy model, but you appear to be unaware of why RFC3460 is badly broken. So yes, you do need to rebuild this.

#4. DOES NOT SUPPORT ECA POLICIES. NEMO USES A CONDITION-ACTION PARADIGM TO EXECUTE ITS POLICY COMMANDS.
It is inconceivable to me how you could disagree with this. From draft-xia-sdnrg-nemo-language-02, Section 5.1: "All the policies follow the same pattern "when <condition>, to execute <action>, with <constraint>", and can be applied to any entity." This is NOT an ECA policy; any useful implementation of CA or ECA allows constraints - EXCEPT RFC3460.


John


On Wed, Jul 15, 2015 at 6:38 PM, Xiayinben <[e-mail]<mailto:[e-mail]>> wrote:
Hi Tina and all,

I do not agree with your word about NEMO/ibnemo in page 8.

Your sentence: “NEMO uses a flatter, simpler object model with fewer objects to represent targets of policy. NEMO does not define a policy model, and does not support ECA policies. NEMO uses a condition-action paradigm to execute its policy commands.”
The truth is that NEMO define an object model and an operation model to represent network abstraction for intent NBI, because NEMO’s target is a simplified network interface(Intent NBI) for application even for ender-user.  policy is only a type of operation in Intent NBI. NEMO doesn’t NEED to define a new policy model, because industry already has well defined policy model(such as RFC 3460). We will not rebuild a wheel.

Your sentence: “IBNemo – 'intent' means I express a desired topology and its properties.”
The truth  is that Ibnemo –‘Intent’ means user simply and intuitively  express demands of network resource and/or network behavior. Topology and its properties are typical type of network resource, so they are part of network intent, but not all.

Best regards,
Yinben
发件人: Supa [mailto:[e-mail]] 代表 Tina TSOU
发送时间: 2015年7月15日 15:42
收件人: Scott O. Bradner; Susan Hares
抄送: [e-mail]<mailto:[e-mail]>
主题: Re: [Supa] SUPA call: 7:30-8:30am, Monday, July 13, San Francisco time

Dear all,

Enclosed pls find the updated slides based on Monday July 13 call discussion, only page 8 added.


Thank you,
Tina

From: Supa [mailto:[e-mail]] On Behalf Of Tina TSOU
Sent: Friday, July 10, 2015 4:31 PM
To: Scott O. Bradner; Susan Hares
Cc: [e-mail]<mailto:[e-mail]>
Subject: Re: [Supa] SUPA call: 7:30-8:30am, Monday, July 13, San Francisco time

Dear Scott and Sue,

Regarding “2.2. Relationship to other WGs (Scott/Sue et al, 5 minutes)” (Times are the ones in the call next week – not the BOF), attached is a quick update based on Jun’s presentation in last BoF, according to the discussion after Dallas.

Feel free to use or update the slides.


Thank you,
Tina

From: Supa [mailto:[e-mail]] On Behalf Of Tina TSOU
Sent: Tuesday, July 07, 2015 7:05 AM
To: [e-mail]<mailto:[e-mail]>
Subject: [Supa] SUPA call: 7:30-8:30am, Monday, July 13, San Francisco time


Dear all,



Below is the draft agenda and logistics for next call during 7:30-8:30am, Monday, July 13, San Francisco time.



Look forward to talking to many of you at the meeting.



----------

1. Note Well, logistics, agenda bashing (chairs, 5 min)



2. Slides Preparation for the BoF (Times are the ones in the call next week – not the BOF)



2.1. Goals, Framework, Examples (Max, Oscar (Luis) et al, 10 minutes)

2.2. Relationship to other WGs (Scott/Sue et al, 5 minutes)

2.3. Information Model Overview (John et al, 5 minutes)

2.4. Data Model Overview (Andy et al, 5 minutes)

2.5. Charter presentation (Dan/Juergen et al, 10 minutes)

2.6. Discussion on whether the progress assessment implies going forward changes to the charter (all, 15 minutes)

2.7. Questions to the audience (chairs, 10 minutes)



-----------





Join WebEx meeting<https://ietf.webex.com/ietf/j.php?MTID=m23e113c8822a0970e0e779383ab09322>

Meeting number: 646 022 302

Meeting password:       2015





Join by phone

1-877-668-4493<tel:[phone]> Call-in toll free number (US/Canada)

1-650-479-3208<tel:[phone]> Call-in toll number (US/Canada)

Access code: 646 022 302

Toll-free calling restrictions<http://www.webex.com/pdf/tollfree_restrictions.pdf>







Can't join the meeting? Contact support.<https://ietf.webex.com/ietf/mc>



IMPORTANT NOTICE: Please note that this WebEx service allows audio and other information sent during the session to be recorded, which may be discoverable in a legal matter. By joining this session, you automatically consent to such recordings. If you do not consent to being recorded, discuss your concerns with the host or do not join the session.





Thank you,

Dan & Tina

_______________________________________________
Supa mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/supa



--
regards,
John