Re: [Teas] Why term transport slice? WG adoption - draft-nsdt-teas-transport-slice-definition

["Rokui, Reza (Nokia - CA/Ottawa)" <[e-mail]>]

Hi Deborah,

>>>>>>> but what does this have to do with TEAS work? If point 1 with Figure 1 is your basis of justification for “transport”, with a segmented view of the network, while it may be correct from a technology view, it is not the basis of TEAS work

One important aspect that I forgot to add to my previous email will hopefully answer your valid question on how transport slice work relates to ACTN.
Each “Transport Slice” (or any other name adopted later) has two characteristics:

  *   Transport slice definition: which as described in my previous email is a set of connectivity among endpoints to satisfy specific SLA/SLO. This is network and technology-agnostics.
  *   Transport slice realization: Describe how the transport slice is implemented in the network. This is technology dependent.

For example suppose you have a transport slice defined as two connections between EP1, EP2 with SLO1. If your network is IP/MPLS with TE enabled, the operator can realize this transport slice using any IETF Service/Paths/Tunnels models (including ACTN).
However, if the operator’s network for this transport slice is a PON network, the realization will be complexly different. However, in both cases, the transport slice definition remains exactly the same but the realization changes.
The mapping between technology-agnostics and technology-specific is done between Transport Slice Controller (TSC) NBI and SBI (See picture below).

In summary to answer your question, a transport slice can be realized by different technologies. One of them can be ACTN as shown in following figure taken from transport slice framework draft (https://tools.ietf.org/html/draft-nsdt-teas-ns-framework-02). This topic was discussed at NSDT in length. Dhruv was involved in this discission.
In other words, the transport slice work done at NSDT will complement IETF ACTN work.


[cid:[e-mail]361BE0]

Reza


From: "BRUNGARD, DEBORAH A" <[e-mail]>
Date: Friday, September 4, 2020 at 11:46 AM
To: Reza Rokui <[e-mail]>
Cc: Kiran Makhijani <[e-mail]>, "[e-mail]" <[e-mail]>, Eric Gray <[e-mail]>, Igor Bryskin <i_bryskin=[e-mail]>, TEAS WG Chairs <[e-mail]>, Vishnu Pavan Beeram <[e-mail]>, TEAS WG <[e-mail]>
Subject: Re: Why term transport slice? [Teas] WG adoption - draft-nsdt-teas-transport-slice-definition

Hi Reza,
(as AD)

Much thanks for your detailed explanation.

I think I’m beginning to see the confusion - my confusion was as Adrian noted - not with the document itself - it is a nicely written document - but what does this have to do with TEAS work? If point 1 with Figure 1 is your basis of justification for “transport”, with a segmented view of the network, while it may be correct from a technology view, it is not the basis of TEAS work. There are many RFCs, especially the ASON work (before TEAS - in CCAMP) will show a different view of the network. Adrian tried to summarize. Especially when proprietary domains are present.

TEAS is responsible for TE architectures with generic applicability - I’ve seen in this thread the desire by some of the authors to include non-TE. I understand the need - but that’s not TEAS. As Adrian noted - is TEAS the proper home?

Re-reading this definition document, I realize TE is not even mentioned except to say this work is different from a TE link and so needs to be defined differently. On this reading, I see the key element is the transport slice controller NBI.

Maybe this author team is not familiar with ccamp, ccamp is already defining a Transport (yes, “Transport”:-)) NBI? And Client- service models. A 3GPP transport NBI can either be done as an applicability or a PS document as part of their work. In ccamp, “transport” slice will not be confusing.

I think it was great for the DT to start sorting out the gap for 3GPP transport work. I’ll talk with the chairs to sort out the best home. While sorting out, please continue discussing and progressing the document. There’s lots of overlap among TEAS and CCAMP participants on the TEAS list.

Deborah

Sent from my iPhone


On Sep 4, 2020, at 10:16 AM, Rokui, Reza (Nokia - CA/Ottawa) <[e-mail]> wrote:
All,

Thanks for all feedbacks.

Let’s step back and address these questions: How and why the authors’ of the draft chose the term “Transport Slice”?

Before formation of the TEAS WG NSDT, there were lots of discussions and drafts to address the role of IETF for network slicing. Those discussion and drafts tried to address the network slicing  from different perspectives but in most cases they had one thing in common, they started by discussion the network slicing but at the end they really meant the Transports portion of the network slice. In other words, although the name of the draft and discussion was network slicing, but they just talked about Transport portion.  In other hand, the term network slice and Transport portion of a network slice were used interchangeably.

After creation of the NSDT, we collectively thoughts that the first order of business is to clarify this. So, the “draft definition” started. The following are the reasons:

Reason 1)
The first reason for this draft is to make very clear distinction between a network slice (defined for example by 3GPP) and transport portion of a network slice.
In our opinion it is essential to make a clear distinction between network slice and transport portion of a network slice. They are NOT the same since a network slice contain the transport portion.
The picture below was outcome of that discussion. In summary, a network slice is an end-to-end context and depends on the used case (i.e 5G, DCI, etc), it might contain a few other components (i.e. RAN, Transport, Core etc.)

<image001.png>

Reason 2)
We just established the fact that an end-to-end network slice is different from transport portion of the network slice. The next question is that what the definition of the Transport portion of a network slice is.
This is fully discussed in draft but in summary the transport portion of a network slice describes the CONNECTIVITY between various endpoints. Our definition is aligned with MEF and 3GPP.

  *   MEF uses the same definition for Transport portion of the network slice”. See Section 5.3 of following white paper

     *   https://wiki.mef.net/display/CESG/Slicing+for+Shared+5G+Fronthaul+and+Backhaul+-+White+Paper<https://urldefense.proofpoint.com/v2/url?u=https-3A__wiki.mef.net_display_CESG_Slicing-2Bfor-2BShared-2B5G-2BFronthaul-2Band-2BBackhaul-2B-2D-2BWhite-2BPaper&d=DwMGaQ&c=LFYZ-o9_HUMeMTSQicvjIg&r=6UhGpW9lwi9dM7jYlxXD8w&m=W4iKXLRn0c02KfMb6XuMzOUPI3bVoc8fMH1OTJcjqgg&s=V0cMMN7woyDsVMTdz_QamIGXjKW8FEqDtZrkpFk8D5M&e=>



  *   This is aligned with 3GPP. See Figure 4.9.3.1 of TR 28.801 and  http://www.3gpp.org/NEWS-EVENTS/3GPP-NEWS/1951-SA5_5G<https://urldefense.proofpoint.com/v2/url?u=http-3A__www.3gpp.org_NEWS-2DEVENTS_3GPP-2DNEWS_1951-2DSA5-5F5G&d=DwMGaQ&c=LFYZ-o9_HUMeMTSQicvjIg&r=6UhGpW9lwi9dM7jYlxXD8w&m=W4iKXLRn0c02KfMb6XuMzOUPI3bVoc8fMH1OTJcjqgg&s=tROrnxxHAgZdPFGiZ62rgSO_iw1Nb82TkIzKwWZkVUY&e=>

According to the picture below,  the reference of transport portion of a network slice  is referred by “Transport network supporting connectivity’
<image002.png>




Reason 3)
The next question is that which term shall be used for ‘Transport portion of an end-to-end network slice”?


  *   MEF uses the term “Transport Slice”. See Figure 17 of following white paper

     *   https://wiki.mef.net/display/CESG/Slicing+for+Shared+5G+Fronthaul+and+Backhaul+-+White+Paper<https://urldefense.proofpoint.com/v2/url?u=https-3A__wiki.mef.net_display_CESG_Slicing-2Bfor-2BShared-2B5G-2BFronthaul-2Band-2BBackhaul-2B-2D-2BWhite-2BPaper&d=DwMGaQ&c=LFYZ-o9_HUMeMTSQicvjIg&r=6UhGpW9lwi9dM7jYlxXD8w&m=W4iKXLRn0c02KfMb6XuMzOUPI3bVoc8fMH1OTJcjqgg&s=V0cMMN7woyDsVMTdz_QamIGXjKW8FEqDtZrkpFk8D5M&e=>



  *   3GPP: See Figure 4.9.3.1 of TR 28.801 and  http://www.3gpp.org/NEWS-EVENTS/3GPP-NEWS/1951-SA5_5G<https://urldefense.proofpoint.com/v2/url?u=http-3A__www.3gpp.org_NEWS-2DEVENTS_3GPP-2DNEWS_1951-2DSA5-5F5G&d=DwMGaQ&c=LFYZ-o9_HUMeMTSQicvjIg&r=6UhGpW9lwi9dM7jYlxXD8w&m=W4iKXLRn0c02KfMb6XuMzOUPI3bVoc8fMH1OTJcjqgg&s=tROrnxxHAgZdPFGiZ62rgSO_iw1Nb82TkIzKwWZkVUY&e=>

They do not directly address the transport portion of a network slice. They do not have a term for this. They mainly address the 5G RAN and 5G Core. As shown in the picture above, the reference of transport is phrase “Transport network supporting connectivity”.



  *   From IETF point of view, these are potential choices for Transport portion of a network slice:

  *   Network slice: This for sure is NO. Reason 1) clearly shows that we shall not use term “Network slice” for transport portion. This is not correct.
  *   Use 3GPP phrase: “Transport network supporting connectivity”
  *   Use the phrase “Transport portion of the Network Slice”
  *   Use term “Transport Network Slice”
  *   Use term “Transport Slice”
  *   Adrian, Igor, Deborah and others, is there any other suggestion? If so, please add

From the above choices, the draft authors uses the term “Transport Slice” but not the “Transport Network Slice” to make sure we implicitly stating that Network Slice and Transport part are different.
Having said that, authors are open to suggestion. Please suggest your term.


Reza