Hi, Tony:
Not got well all your comments :(
Wish the explanation in previous mail to Les can change your impression to this 
draft a bit.

Best Regards.

Aijun Wang
China Telecom

> On Feb 22, 2019, at 14:42, Tony Przygienda <[email protected]> wrote:
> 
> I read it during some mildly boring phone conference here ;-) I do 
> (unsuprisingly) tend to agree with Mr. Ginsberg here. Putting originator on 
> prefix, well, ok, I understand how people got there instead of having a clean 
> router LSA with its loopback hanging of it (which would be IMO the right 
> abstraction but alas, IP decided originally that interfaces have addresses 
> and routers, well, don't. Overall not the best architecture remedied by 
> loopback hacks since forever then. I digress ...). So putting router-id on 
> type-3 loopback serves its purpose of "how do I get to this node across 
> topology abstractions [areas]". Allows central entities (controllers and 
> such) to get to every node in the topology without violating abstractions 
> (too much ;-). Cleaner solutions would be thinkable but more complex 
> obviously. But coming back to this draft, alas, trying to build the whole 
> topology distribution tails-up, i..e. redistribute topology view behind 
> prefixes which are really leafs-of-topology strikes me just another confused 
> slippery slope just like re-encoding one protocol's native formats into 
> another protocol :-P ... And if the authors suffer from mild bout of 
> over-creativity I suggest to split that part into an informational draft ... 
> 
> --- tony 
> 
>> On Thu, Feb 21, 2019 at 9:56 PM Les Ginsberg (ginsberg) <[email protected]> 
>> wrote:
>> Aijun –
>> 
>>  
>> 
>> Controllers already have a reliable way to learn topology information for 
>> all areas.
>> 
>> There is therefore no need for the topology discovery solution you propose – 
>> and all the more so because your proposal does not work in all cases and you 
>> have no definition of how a controller could tell when the information can 
>> be trusted and when it can’t.
>> 
>>  
>> 
>> The only thing which is needed is to define a way to identify the source 
>> router-id of prefixes which are leaked between areas – which is what I have 
>> asked you to limit the draft to do.
>> 
>>  
>> 
>> The mention of ELC/ERLD/MSD in your draft is spurious since you actually 
>> haven’t defined any way to advertise that information between areas (nor do 
>> I want you to do so).  It should be removed.
>> 
>>  
>> 
>> I say again, this draft in its current form is not ready to be adopted.
>> 
>>  
>> 
>>    Les
>> 
>>  
>> 
>>  
>> 
>> From: Aijun Wang <[email protected]> 
>> Sent: Thursday, February 21, 2019 3:27 PM
>> To: John E Drake <[email protected]>
>> Cc: Les Ginsberg (ginsberg) <[email protected]>; Acee Lindem (acee) 
>> <[email protected]>; [email protected]
>> Subject: Re: [Lsr] Working Group Adoption Poll for "OSPF Extension for 
>> Prefix Originator" - draft-wang-lsr-ospf-prefix-originator-ext-01
>> 
>>  
>> 
>> Hi, John:
>> 
>>  
>> 
>> Thanks for your review and comments.
>> 
>> The use cases and original thought in this draft are different from that  
>> described in RFC7794. We have pointed out that RFC7794 has the similar 
>> extension for ISIS and indicated that the extension for ISIS can also be 
>> used in the use cases described in our draft. What’ other content do you 
>> think it is needed further?
>> 
>> RFC 7770  solves mainly the advertising of router’s capabilities, it 
>> shouldn’t be used for transmitting the information about the prefixes.
>> 
>>  
>> 
>> Best Regards.
>> 
>>  
>> 
>> Aijun Wang
>> 
>> China Telecom
>> 
>> 
>> On Feb 21, 2019, at 23:41, John E Drake 
>> <[email protected]> wrote:
>> 
>> Hi,
>> 
>>  
>> 
>> I agree with Les.  I think the draft should be recast to indicate that it is 
>> providing OSPF parity with RFC 7794.
>> 
>> Can’t topology discovery be done using RFC 7770?
>> 
>>  
>> 
>> Yours Irrespectively,
>> 
>>  
>> 
>> John
>> 
>>  
>> 
>> From: Lsr <[email protected]> On Behalf Of Les Ginsberg (ginsberg)
>> Sent: Monday, February 18, 2019 8:22 AM
>> To: Acee Lindem (acee) <[email protected]>; [email protected]
>> Subject: Re: [Lsr] Working Group Adoption Poll for "OSPF Extension for 
>> Prefix Originator" - draft-wang-lsr-ospf-prefix-originator-ext-01
>> 
>>  
>> 
>> To the extent that the draft defines functionality equivalent to that 
>> defined in IS-IS RFC 7794 – specifically a means to advertise the source 
>> router-id of a given advertisement – it defines a necessary and useful 
>> extension to the OSPF protocol – and I support that work.
>> 
>>  
>> 
>> However, in its current form the draft discusses use of this mechanism for 
>> inter-area topology discovery. This idea is seriously flawed – as has been 
>> discussed extensively on the WG list.
>> 
>> The draft also discusses uses cases related to ERLD, the direction for which 
>> is very much uncertain at this time.
>> 
>>  
>> 
>> I therefore feel that the current content of the draft is not what I would 
>> expect to see approved by the WG as an RFC and therefore have significant 
>> reservations about moving forward with the existing content.
>> 
>>  
>> 
>> I do want to see a draft addressing the source router-id advertisement gap 
>> move forward – and if this draft is reduced to focus on that then I can 
>> enthusiastically support adoption – but in its current form I cannot 
>> indicate support.
>> 
>>  
>> 
>>    Les
>> 
>>  
>> 
>>  
>> 
>> From: Lsr <[email protected]> On Behalf Of Acee Lindem (acee)
>> Sent: Wednesday, February 13, 2019 5:26 AM
>> To: [email protected]
>> Subject: [Lsr] Working Group Adoption Poll for "OSPF Extension for Prefix 
>> Originator" - draft-wang-lsr-ospf-prefix-originator-ext-01
>> 
>>  
>> 
>> This begins a two week adoption poll for the subject draft. Please send your 
>> comments to this list before 12:00 AM UTC on Thursday, February 28th, 2019.
>> 
>>  
>> 
>> All authors have responded to the IPR poll and there is one  
>> https://datatracker.ietf.org/ipr/search/?submit=draft&id=draft-wang-lsr-ospf-prefix-originator-ext
>> 
>> It is listed multiple times but references the same CN201810650141.
>> 
>>  
>> 
>> Thanks,
>> 
>> Acee
>> 
>>  
>> 
>> _______________________________________________
>> Lsr mailing list
>> [email protected]
>> https://www.ietf.org/mailman/listinfo/lsr
>> 
>> _______________________________________________
>> Lsr mailing list
>> [email protected]
>> https://www.ietf.org/mailman/listinfo/lsr
> _______________________________________________
> Lsr mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/lsr
_______________________________________________
Lsr mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/lsr

Reply via email to