Re: [Lime-oam-model] OAM Model analisys. First cut

[Qin Wu <[e-mail]>]

When is the next LIME design team meeting? It looks we have sorted out a lot of thing.

-Qin
-----邮件原件-----
发件人: Lime-oam-model [mailto:[e-mail]] 代表 Tom Taylor
发送时间: 2015年2月9日 5:28
收件人: [e-mail]
主题: Re: [Lime-oam-model] OAM Model analisys. First cut

Could we get away from the coded terminology and use something more intuitive? Obviously we're venturing into management architecture here, but presumably what we need to distinguish to start with (I'm sure there's more) are:

-- defect detection points -- talk to the reporting point
-- defect reporting points -- talk to the manager
-- test injection points -- talk to the manager
-- test responding points -- talk to the manager
-- target trail starting point -- obeys the test injection point?
-- target trail ending point -- obeys the test injection point?

These may be collocated in any combination. For instance, for BFD, I believe the test injection point and test responding point are collocated.

Tom

On 08/02/2015 2:30 PM, Gregory Mirsky wrote:
> Hi Deepak,
> these are very fine questions.
> I’m not sure that MP (MEP/MIP) has much value outside of OAM model but for our effort to build OAM YANG data model, whether it will be one common for all IETF OAMs or several, it is one of the cornerstones. I don’t have answers to all the questions, at least not yet. And your ideas, suggestions, as of all of our DT, very much welcome and urgently needed.
>
>                  Regards,
>                                  Greg
>
> From: Deepak Kumar (dekumar) [mailto:[e-mail]]
> Sent: Saturday, February 07, 2015 9:29 AM
> To: Gregory Mirsky; Qin Wu; [e-mail]
> Subject: Re: [Lime-oam-model] OAM Model analisys. First cut
>
> Hi Greg,
>
> As MEP and MIP are implicit in IP how does defining a Yang model to fill this implicit MEP/MIP specific information changes IP OAM protocol as we are not carrying this MEP/MIP information over the protocol or defining new RPC.
> Again this is not OAM specific information but Management specific information for operator(s).
>
> We need this Test point(s) for Yang Application or Server to efficiently debug the problem and Also these Test points give information regarding the underlying Server layer, so it aid application to go down layer one at a time to isolate Fault.
>
> If we don't provide these Test points and intelligence provided by network, then how does Yang effort will be different than snmp mibs currently present except some perfomance improvements.
>
> Thanks,
> Deepak
>
> From: Gregory Mirsky 
> <[e-mail]<mailto:[e-mail]>>
> Date: Friday, February 6, 2015 10:39 PM
> To: Qin Wu <[e-mail]<mailto:[e-mail]>>, 
> "[e-mail]<mailto:[e-mail]>" 
> <[e-mail]<mailto:[e-mail]>>
> Subject: Re: [Lime-oam-model] OAM Model analisys. First cut
>
> Hi Qin,
> as I understand, LIME WG is not to define new OAM but to create YANG data model for existing. You’re proposing work on introducing new elements into IP and IP/MPLS OAM models. That may be useful but I’m not sure that that is in LIME WG charter. Hope our WG chairs will direct us.
>
>                  Regards,
>                                  Greg
>
> From: Qin Wu [mailto:[e-mail]]
> Sent: Friday, February 06, 2015 10:35 PM
> To: Gregory Mirsky; 
> [e-mail]<mailto:[e-mail]>
> Subject: RE: OAM Model analisys. First cut
>
> Why MD/MA or MEG should be coupled with Ethernet OAM or MPLS-TP? They are just some terminology. We can make these term more generialized.
> What’s the harm to apply MD/MA or MAG to IP and IP/MPLS, I see the 
> benefit Of using MD/MA or MAG to provide consistent reporting and 
> representation, Especially in the end to end path diagnose case.
>
> Also LIME’s goal is to provide consistent reporting, representation and configuration.
>
> Regards!
> -Qin
> 发件人: Gregory Mirsky [mailto:[e-mail]]
> 发送时间: 2015年2月7日 14:27
> 收件人: Qin Wu; [e-mail]<mailto:[e-mail]>
> 主题: RE: OAM Model analisys. First cut
>
> Hi Qin,
> could you point to where in IP and IP/MPLS we can find definition of elements analogous to MD/MA or MEG; MP (both MEP and MIP)? In the table I’ve marked all of them as absent in IP and IP/MPLS OAM? Do you think I’ve missed them?
>
>                  Regards,
>                                  Greg
>
> From: Qin Wu [mailto:[e-mail]]
> Sent: Friday, February 06, 2015 10:11 PM
> To: Gregory Mirsky; 
> [e-mail]<mailto:[e-mail]>
> Subject: RE: OAM Model analisys. First cut
>
> Here is my understanding
> IP is connection less but IP can use upper layer protocol to setup connection with other hosts, even though it is communicating over connection less network. The upper layer can be connectionless or connection-oriented.
> Ethernet is also connectionless. Ethernet can use upper layer protocol or is encapsulated into connection oriented network technology to set up connections with other hosts, even though it is communicating over connectionless network and data-link layer protocols.
> It does not matter whether connectionless or connection-oriented.
>
> Whether connectionless or connection-oriented, the common elements they have are fault domain, test points, addressing, layering.
>
> -Qin
> 发件人: Gregory Mirsky [mailto:[e-mail]]
> 发送时间: 2015年2月7日 0:40
> 收件人: Qin Wu; [e-mail]<mailto:[e-mail]>
> 主题: RE: OAM Model analisys. First cut
>
> Hi Qin,
> IP and IP/MPLS networks are distinctly connectionless and their OAM models, in my view, share very little common elements with Ethernet OAM, which primarily serves transport, i.e. connection-oriented, network, as well as MPLS-TP.
>
>                  Regards,
>                                  Greg
>
> From: Qin Wu [mailto:[e-mail]]
> Sent: Wednesday, February 04, 2015 11:34 PM
> To: Gregory Mirsky; 
> [e-mail]<mailto:[e-mail]>
> Subject: RE: OAM Model analisys. First cut
>
> Sure, IP based OAM model and CFM based OAM model have different 
> addressing, CFM is originally designed for Ethernet technology. but CFM based OAM model support both connection less and connection oriented communication.
> Becos Ethernet technologies support both connection oriented and connection less.
>
> Regards!
> -Qin
> 发件人: Gregory Mirsky [mailto:[e-mail]]
> 发送时间: 2015年2月5日 14:59
> 收件人: Qin Wu; [e-mail]<mailto:[e-mail]>
> 主题: RE: OAM Model analisys. First cut
>
> Hi Qin,
> there’s no surprise that CFM and Y.1731 follow the same model. Nor that MEF documents you’ve referenced so close to Y.1731, since Carrier Ethernet is, in fact and for practical purposes, Y.1731. And for those who followed MPLS-TP JWT strong resemblance of Y.1731 is not surprise either.
> What I think we’re trying to establish is whether there is enough commonality between IP/IP-based OAM models and CFM/Y.1731/MPLS-TP/TRILL OAM.
>
>                  Regards,
>                                  Greg
>
> From: Qin Wu [mailto:[e-mail]]
> Sent: Wednesday, February 04, 2015 10:50 PM
> To: Gregory Mirsky; 
> [e-mail]<mailto:[e-mail]>
> Subject: RE: OAM Model analisys. First cut
>
> Here is my point.
> Using any model from other SDOs as basis doesn’t mean we step into other SDO territory.
>  From Several reference document I listed, we can see what are LIME related work?
> What commonality are they sharing?
> e.g., IEEE 802.1Q based model is similar to Y.1731 based model MEF 38 
> and MEF39 use 802.1Q based model, i.e., CFM model.
> RFC6371 uses Y.1731 based model as basis.
> We should not deny this fact, we should not neglect non-IETF work by worrying about stepping into other SDO territory.
>
> Regards!
> -Qin
> 发件人: Gregory Mirsky [mailto:[e-mail]]
> 发送时间: 2015年2月5日 14:30
> 收件人: Qin Wu; [e-mail]<mailto:[e-mail]>
> 主题: RE: OAM Model analisys. First cut
>
> Hi Qin,
> many thanks for your suggestion. I think that we should define the scope of our discussion. I believe that non-IETF technologies should be out of scope and only IETF ones analyzed. That is why I’ve included only IP, IP/MPLS, MPLS-TP and TRILL. If we to use the list of references, then we may be stepping into IEEE, ITU-T and MEF territory. Hope our WG chairs can clarify that.
>
>                  Regards,
>                                  Greg
>
> From: Qin Wu [mailto:[e-mail]]
> Sent: Wednesday, February 04, 2015 7:15 PM
> To: Gregory Mirsky; 
> [e-mail]<mailto:[e-mail]>
> Subject: RE: OAM Model analisys. First cut
>
> One thing I am surprising is there is no references have been provided for this analysis.
> If we have all the referenced document put together, it will be easy 
> to help us to find what are common things in these various OAM technologies.
> e.g., some references I can provide include:
> a. IEEE 802.1Q
> b. ITU-T Y.1731
> c.  MPLS-TP OAM model in the section 4 of RFC6371 d. MEF-38 Service 
> OAM Fault Management YANG Modules Technical Specification e. MEF-39 
> Service OAM Performance Monitoring YANG Module Technical Specification
>
> -Qin
> 发件人: Gregory Mirsky [mailto:[e-mail]]
> 发送时间: 2015年2月5日 3:55
> 收件人: Qin Wu; [e-mail]<mailto:[e-mail]>
> 主题: RE: OAM Model analisys. First cut
>
> Hi Qin,
> I believe that LIME charter is not to develop new OAM tools but to work on OAM data model that is relevant to all or some OAM models. If some OAM models do not have some of functions we find in other models, then we can flag that and discuss but in other group. The table you find in my writing is only start of what OAM model analysis may look like. I’d like the team to review it and help with comments and suggestions to extend the table. Then we can come to informed decision on what is common among IETF OAM models and discuss how useful data model of that common set may be.
>
>                  Regards,
>                                  Greg
>
> From: Qin Wu [mailto:[e-mail]]
> Sent: Tuesday, February 03, 2015 6:03 PM
> To: Gregory Mirsky; 
> [e-mail]<mailto:[e-mail]>
> Subject: RE: OAM Model analisys. First cut
>
> Greg:
> What conclusion do you make from this analysis? Do you want to make BDI and FDI being supported as common OAM function? Or do you want to make linktrace and loop back layer independent? Aren’t linktrace and loopback Ethernet specific functions?
> I think the example you give in the table 1 gets complete alignment with the Table 4 in the section 5.2 of RFC7276.
> [cid:image001.jpg@01D0425D.C4218FF0]
> That is to say Continuity Check, Connectivity Verification, Path Discovery, Performance measurement are common OAM functions for the all the OAM technologies.
> Besides Common OAM functions, we also have fault domain, test point, 
> addressing, technology type, sub technology type, specific layer. They are all common things for all the OAM technologies.
>
> Regards!
> -Qin
> 发件人: Lime-oam-model [mailto:[e-mail]] 代表 
> Gregory Mirsky
> 发送时间: 2015年2月4日 9:22
> 收件人: [e-mail]<mailto:[e-mail]>
> 主题: [Lime-oam-model] OAM Model analisys. First cut
>
> Dear All,
> apologies for belated update. Attached please find first cut of the proposal for approach to OAM Model comparison and example of it being used.
>
> Your questions, comments, suggestions always welcome and greatly appreciated.
>
>                  Regards,
>                                  Greg
>
>
>
> _______________________________________________
> Lime-oam-model mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/lime-oam-model
>

_______________________________________________
Lime-oam-model mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/lime-oam-model