Re: [CCAMP] 答复: WG adoption poll on draft-zheng-ccamp-yang-otn-slicing-03

[Igor Bryskin <[e-mail]>]

Hi Adrian, 
Thanks for your inside.
Firstly, I still can't see a daylight between L1VPN and OTN slice. I'd also like to hear from the authors something more than L1VPN is one way to implement OTN slice. What would be other ways? And what about  L0? Would we also need Och slicing, microwave slicing, etc.? I think such a discussion would be welcome addition to the draft.
Secondly, we put a lot of effort into L1VPNs - we had a dedicated WG, which successfully fulfilled its charter and was closed with honors. Looking back though, I hope you'd agree that the level of L1VPN success/deployments, to put it mildly, has not been as high, as we had hoped it would be.  In your opinion why is that? Could it be that because over the years folks realized it is far simpler and straightforward (yet sufficient) to plan and use OTN resources in a shared manner, providing simple p2p connectivity services, rather than breaking the resources into VPNs and providing more complex L1VPN services?
Finally, I have a hypothetical question for you. Imagine that the IP layer data plane was naturally sliced as OTN data plane. For example, imgine IP layer where the only way to achieve any connectivity is via RSVP TE controlled LSPs, which have impenetrable walls separating them from each other (regardless whether you want such walls or not). Do you think we would be working on slicing in such IP?
Thanks,Igor

Sent from Yahoo Mail on Android 
 
  On Mon, Jan 17, 2022 at 5:54 AM, Daniele Ceccarelli<[e-mail]> wrote:   <!--#yiv4373471841 filtered {}#yiv4373471841 filtered {}#yiv4373471841 filtered {}#yiv4373471841 filtered {}#yiv4373471841 filtered {}#yiv4373471841 filtered {}#yiv4373471841 p.yiv4373471841MsoNormal, #yiv4373471841 li.yiv4373471841MsoNormal, #yiv4373471841 div.yiv4373471841MsoNormal {margin:0cm;font-size:12.0pt;font-family:"Times New Roman", serif;}#yiv4373471841 a:link, #yiv4373471841 span.yiv4373471841MsoHyperlink {color:blue;text-decoration:underline;}#yiv4373471841 span.yiv4373471841EmailStyle18 {font-family:"Calibri", sans-serif;color:windowtext;}#yiv4373471841 .yiv4373471841MsoChpDefault {font-family:"Calibri", sans-serif;}#yiv4373471841 filtered {}#yiv4373471841 div.yiv4373471841WordSection1 {}-->
Hi Aihua,
 
  
 
The distinction between realization and intention makes sense. I believe that from the realization point of view they are identical (control plane excluded) but from an intent point of view there is a difference.
 
  
 
Cheers,
Daniele  
 
  
 
From: Aihua Guo <[e-mail]>
Sent: den 14 januari 2022 23:20
To: [e-mail]
Cc: Daniele Ceccarelli <[e-mail]>; Igor Bryskin <[e-mail]>; Fatai Zhang <zhangfatai=[e-mail]>; CCAMP <[e-mail]>
Subject: Re: [CCAMP] 答复: WG adoption poll on draft-zheng-ccamp-yang-otn-slicing-03
 
  
 
Hi Adrian, Daniele and Igor,
 
  
 
IMHO an OTN slice, like an IETF network slice, is used to represent an intent from a (internal or external) customer's view. This intent includes a combination of resource requirements which can be expressed by connectivities or topologies, along with the common and OTN technology-specific SLOs. Customers usually do not care how an OTN slice is realized at the control plane and/or data plane, and VPN is one of the possible realizations. 
 
  
 
To summarize, an L1VPN with QoS parameters can be considered as a possible realization of an OTN slice. 
 
  
 
Also, if I understand it correctly, L1VPN is implemented using the control plane per rfc4878 and is not targeted as an NBI model. 
 
  
 
Thanks,
 
Aihua
 
  
 
On Fri, Jan 14, 2022 at 11:57 AM Adrian Farrel <[e-mail]> wrote:
 

I think there is a difference between a L3/L2 VPN and a L1VPN that is to do with the nature of connectivity in the underlying network. A packet VPN is expressed as a set of edge points and the edge connectivity capacities, but not the edge-to-edge capacities. A L1VPN (at least for some L1 technologies – but not all?) has to be expressed in terms of edge-to-edge capacities because you cannot do “on-demand” traffic routing within the network – the flow markings are at a macro level.
 
 
 
>From that perspective, I’d agree that L1VPN is close to L1 slicing. It would be good to hear the authors spell out the differences that they obviously have in mind.
 
 
 
But, not only did Daniele ask his question in TEAS, he also go an answer😊. There are service models in which a network slice may look like a “VPN with QoS” – those are the enhanced VPN cases. But there are also service models where the edge-to-edge flows are spelled out in more detail than in a VPN.
 
 
 
Ciao,
 
Adrian
 
 
 
From: Daniele Ceccarelli <[e-mail]>
Sent: 14 January 2022 16:14
To: Igor Bryskin <[e-mail]>; zhangfatai=[e-mail];[e-mail]; 'CCAMP' <[e-mail]>
Subject: RE: [CCAMP] 答复: WG adoption poll on draft-zheng-ccamp-yang-otn-slicing-03
 
 
 
Indeed a very good question…which is similar to the more generic question of what is the difference between a LxVPN with QoS (e.g a L3VPN bound to a TE tunnel) and a slice which I tried to ask in TEAS (but that’s a different story).
 
Let’s focus on the difference between L1VPN and OTN slice here.
 
 
 
BR
Daniele 
 
 
 
From: CCAMP <[e-mail]>On Behalf Of Igor Bryskin
Sent: den 14 januari 2022 09:34
To: zhangfatai=[e-mail]; Fatai Zhang <zhangfatai=[e-mail]>;[e-mail]; 'Fatai Zhang' <zhangfatai=[e-mail]>; 'CCAMP' <[e-mail]>
Subject: Re: [CCAMP] 答复: WG adoption poll on draft-zheng-ccamp-yang-otn-slicing-03
 
 
 
Hi Fatai,
 
 
 
I like that you and Adrian like my style, but I did ask questions before the last IETF. The basic one is this:
 
When I explain my wife what IETF network slice is, I say: Remember, we have this thing called IP VPN?  Network slice is roughly IP VPN with QOS parameters. 
 
 
 
Considering that L1VPN cannot be without QOS, my question to the authors what is the difference between L1VPN and OTN slice?
 
 
 
Cheers,
 
Igor
 
Sent from Yahoo Mail on Android
 
 
 

On Fri, Jan 14, 2022 at 2:00 AM, Fatai Zhang
 
<zhangfatai=[e-mail]> wrote:
 
_______________________________________________
CCAMP mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/ccamp
 

_______________________________________________
CCAMP mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/ccamp