Re: [Ila] review comments on ] draft-ietf-dmm-srv6-mobile-uplane-00.txt

[Satoru Matsushima <[e-mail]>]

Hello Uma,

> 
> When it comes to service function type UPF, you name it. Following draft exhibits how service chain can be done by SRv6:
> https://tools.ietf.org/html/draft-xuclad-spring-sr-service-chaining-00
> 
> [Uma]: I presume this is on N6 interface once de-capsulation is done at eventual UPF.  So can I say this is one more alternative to NSH ??

SRv6 can be a SFP in terms of SFC architecture. SRH is able to bring NSH in a TLV. See https://tools.ietf.org/html/draft-ietf-6man-segment-routing-header-08#section-3.1.6



>                Do you see any relevance of this in any other interface?
> 

I’ve moved it to another thread. I’d like to discuss it in that thread.


>> Also you show IPv6/SRv6 nodes in those slides. Are the UPFs ‘overlaid’ on IPv6/SRv6 nodes?
>> Are these UPFs VNFs? Or are UPFs implemented on IPv6/SRv6 nodes?
>> 
> 
> When you see UPF specifically it should be controlled by SMF through N4, they are not the UPFs.
> But you might see them as UPFs if a SMF doesn’t control them directly but the SMF can put the sessions to it through some other means.
> 
> [Uma]: Didn't quite understand. Are you referring southbound interface like PCEP here?

It looks same question from Behcet. Let me think just an example, if a SMF sees an IPv6 address as an UPF address, is actually an IPv6 segment ID of a TE path through several IPv6 routers and links, a southbound could be PCEP but not limited. BGP-LS should work to disseminate that segment and FPC may work to disclose it to the SMF and the TE path would be attached mobility sessions by the SMF as if it is an UPF. 

Cheers,
--satoru