Re: [Lsr] Erik Kline's Discuss on draft-ietf-lsr-isis-srv6-extensions-14: (with DISCUSS)

[Peter Psenak <[e-mail]>]

Hi Ben,

please see inline:

On 22/05/2021 00:57, Benjamin Kaduk wrote:
> On Thu, May 20, 2021 at 10:59:23AM -0400, Joel Halpern Direct wrote:
>> None of the cases you described are used for routing.
>>
>> And advertising information for which we do not know the use seems a bad
>> idea.
> 
> At least, not something that we should feel strongly obligated to put in a
> Standards-Track specification.
> 
>> The abstraction that lets us talk about func bits and arg bits is nice.
> 
> It is certainly convenient, and probably even worth having written down.
> 
>>    But in fact, the operational structures do not depend upon that.
> 
> I am inclined to agree.  Pulling in a quote from upthread: "the information
> comes from the router, that is where the SRv6 SID is allocated."  The SID
> being *allocated by* the router that "owns" the IPv6 block is key to how I
> came to reconcile myself with RFC 8986.  The router can allocate and put
> structure in its own allocations if it wants to, so long as the protocol
> itself doesn't require it, and things mostly remain fine.
> 
> It's also mostly fine if the router (or other entity) allocating the SID
> values and putting structure on them wants to go and tell somebody else
> what it did and how.

above is exactly what the SID Structure TLV was defined for.

> 
> However, if we go and define in a standards-track document the way to tell
> somebody else how it was done, and the mechanism to do so enforces a
> particular structure on the allocation, that becomes problematic for me.

I don't see where the ISIS SRv6 draft enforces any particular structure 
on the SID allocation.

All that the draft does is that it provides a mechanism to advertise the 
SID structure, if the SID was allocated using the structure defined in 
RFC 8986. On top of that, the SID Structure TLV is optional, please see 
the first sentence of section 9.

So there is absolutely no enforcement in this draft in terms of the 
structure of the SID allocation.

   > I think a lot of my unease with this mechanism would go away if it 
was not
> described or implied that this is the only way to have structure to a SID
> and the option was open for the router to allocate its SIDs (and
> communicate about them) in some other way.

the draft does not say anything about this being the only way to have 
structure associated with the SID. It defines the TLV for the only 
structure that RFC 8986 defined. That's all.

thanks,
Peter



> 
> (That said, I still think it would be better done in a different document
> than this one.)
> 
> -Ben
> 
>> I really think removing section 9 would improve the document and operations.
>>
>> Yours,
>> Joel
>>
>> On 5/20/2021 10:10 AM, Peter Psenak wrote:
>>> Joel,
>>>
>>> On 20/05/2021 15:59, Joel M. Halpern wrote:
>>>> I have been watching this debate, and I am left with the impression that
>>>> the information being defined in section 9 of this draft is simply not
>>>> useful for routing.   It confuses operational information with routing
>>>> information.  Given taht the information has to come from somewhere
>>>> outside the router anyway,
>>>
>>> the information comes from the router, that is where the SRv6 SID is
>>> allocated.
>>>
>>> Similar to a prefix that is configured on the router and is advertised
>>> together with some additional attributes (e.g. tag). These attributes
>>> may or may not be used for routing.
>>>
>>> thanks,
>>> Peter
>>>
>>>
>>>
>>>> iand that it is not going to be consumed by
>>>> the routers who receive the advertisements, why is it here?
>>>>
>>>> Yours,
>>>> Joel
>>>>
>>>> On 5/20/2021 7:49 AM, Peter Psenak wrote:
>>>>> Hi Erik,
>>>>>
>>>>> thanks for your comment, please see inline:
>>>>>
>>>>> On 19/05/2021 03:58, Erik Kline via Datatracker wrote:
>>>>>> Erik Kline has entered the following ballot position for
>>>>>> draft-ietf-lsr-isis-srv6-extensions-14: Discuss
>>>>>>
>>>>>> When responding, please keep the subject line intact and reply to all
>>>>>> email addresses included in the To and CC lines. (Feel free to cut this
>>>>>> introductory paragraph, however.)
>>>>>>
>>>>>>
>>>>>> Please refer to
>>>>>> https://www.ietf.org/iesg/statement/discuss-criteria.html
>>>>>> for more information about DISCUSS and COMMENT positions.
>>>>>>
>>>>>>
>>>>>> The document, along with other ballot positions, can be found here:
>>>>>> https://datatracker.ietf.org/doc/draft-ietf-lsr-isis-srv6-extensions/
>>>>>>
>>>>>>
>>>>>>
>>>>>> ----------------------------------------------------------------------
>>>>>> DISCUSS:
>>>>>> ----------------------------------------------------------------------
>>>>>>
>>>>>> [ section 9 ]
>>>>>>
>>>>>> * I share the concerns of several of the others here about SRv6 SIDs
>>>>>> being
>>>>>>      claimed to be IPv6 addresses but kinda not really being IPv6
>>>>>> addresses
>>>>>>      if their internal structure is exposed outside of the given SR
>>>>>> router.
>>>>>
>>>>>
>>>>> SRv6 SIDs are indeed IPv6 addresses. RFC8986 introduced the SRv6 SID
>>>>> structure. It also goes into allocation of SIDs where it describes the
>>>>> carving out of the Block for SRv6 SIDs in the domain, followed by the
>>>>> allocation of SRv6 Locators to the nodes in the domain. Then the node
>>>>> allocates the function part when instantiating the SID - and all of this
>>>>> is signaled via control plane protocols. This is all exposed and know to
>>>>> the operator who determines the addressing scheme.
>>>>>
>>>>>
>>>>>>
>>>>>>      If "[i]t's usage is outside of the scope of this document", can
>>>>>> this be
>>>>>>      removed for now, and maybe take up the issue at some point in the
>>>>>> future
>>>>>>      by which time a motivating use case might have presented itself?
>>>>>
>>>>>
>>>>> The use-cases have not been described in this document since they were
>>>>> out of the scope of the ISIS protocol operations. Some of the use-cases
>>>>> discussed have been :
>>>>>
>>>>> - automation and verification of blocks/locators and setup of filtering
>>>>> for them at SR domain boundaries
>>>>>
>>>>> - validation of SRv6 SIDs being instantiated and advertised via IGP;
>>>>> these can be learnt by apps/controllers via BGP-LS and then monitored
>>>>> for conformance to the addressing rules set by the operator.
>>>>>
>>>>> - verification and even determination of summary routes to be used for
>>>>> covering the SRv6 Locators and SIDs.
>>>>>
>>>>> There may be other use-cases that may be operator or vendor specific.
>>>>> The use-cases are not within the scope of ISIS protocol extensions and
>>>>> are either operational or implementation specific – hence we said it was
>>>>> out of the scope of this document.
>>>>>
>>>>> If you feel adding these to the document may help to clear your
>>>>> concerns, I can certainly add them.
>>>>>
>>>>> thanks,
>>>>> Peter
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Lsr mailing list
>>>>> [e-mail]
>>>>> https://www.ietf.org/mailman/listinfo/lsr
>>>>
>>>>
>>>
>>
> 
>