Re: [Teas] Customer control of inclusion, policy/intent, etc. [Was: Moving forward with draft-ietf-teas-ietf-network-slices]

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

Thanks Adrian.  That all sounds reasonable to me.
Yours,
Joel

On 5/23/2021 12:59 PM, Adrian Farrel wrote:
> Hi,
> 
> Tarek raised an issue a couple of weeks back.
> 
> It’s about how a network slice customer can request constraints on the 
> resources used in the provider network to realise the slice.
> 
> I believe that, in the context of this high-level document we already 
> have this covered in the section on Service Level Expectations (4.1.2). 
> Obviously, we can edit that text, but it seems to me that this type of 
> request made to the provider is unverifiable by the customer.
> 
> Tarek is right that, when specifying the NBI, this sort of thing needs 
> to be included. I would suggest that, unlike the SLOs, the SLEs will 
> take quite a bit of careful specification. It may be that the SLEs 
> should be specified individually in their own modules that augment the 
> base NBI spec.
> 
> Igor and Joel also had a conversation on this, and I think the 
> conclusion was also that SLEs would normally be used for all such 
> controls (don’t send my traffic through country X, don’t use devices 
> made by company Y, don’t use services provided on DC Z). If, however, 
> the provider chooses to supply additional information about how the 
> underlying network is designed, the customer might operate on that to 
> specify constraints. My understanding of the relationship between 
> provider and an external customer is that the provider is not likely to 
> supply that information in an operational context. That might be 
> different with an internal customer.
> 
> Unless, anyone feels strongly, I propose that we move that discussion to 
> a different thread and a different document.
> 
> Best,
> 
> Adrian
> 
> *From:*Tarek Saad <[e-mail]>
> *Sent:* 07 May 2021 18:46
> *To:* John E Drake <jdrake=[e-mail]>; Igor Bryskin 
> <i_bryskin=[e-mail]>; Igor Bryskin 
> <[e-mail]>; Loa Andersson <[e-mail]>; [e-mail]; 
> [e-mail]; [e-mail]; Oscar González de Dios 
> <[e-mail]>
> *Subject:* Re: [Teas] Moving forward with 
> draft-ietf-teas-ietf-network-slices
> 
> A customer may want to have a say in defining a policy on top of the 
> route that a NS connection can take. In absence of a common language to 
> express such (inclusion/exclusion) policy/intent, does it suffice to 
> express this in terms of geographies (GPS locations, countries, cities, 
> roads), srlgs (srlgs may not have global meaning), diversity needs (from 
> other connections), etc.? Alternatively, does it help if the NBI request 
> expresses such inclusions/exclusions policy using/referencing a 
> “customer-centric” topology that a provider may have furnished earlier 
> to the customer, and which the provider can readily map to their network?
> 
> Regards,
> 
> Tarek
> 
> 
> _______________________________________________
> Teas mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/teas
>