Re: [Teas] Various terms for transport portion of a network slicing

[Adrian Farrel <[e-mail]>]

I think this is good progress, Reza. Thanks for working on it and reporting it.

 

I could live with “IETF Network Slice”.

 

In your table for the entry on “Network Slice” you say:

> Since multiple connections are part of a single "Network Slice", it is not a good 

> idea to call each of these connections "Network slice".

I find this text to be very important because I think it underpins very much of the confusion in the discussions. For my part, I don’t want to apply a name to any single “connection” but want a term that describes the partitioning of underlay network resources for use by one or more overlay connections.

 

Cheers,

Adrian

 

From: Teas <[e-mail]> On Behalf Of Rokui, Reza (Nokia - CA/Ottawa)
Sent: 29 September 2020 19:42
To: Lou Berger <[e-mail]>; TEAS WG <[e-mail]>; Vishnu Pavan Beeram <[e-mail]>; BRUNGARD, DEBORAH A <[e-mail]>; [e-mail]
Cc: Rokui, Reza (Nokia - CA/Ottawa) <[e-mail]>
Subject: [Teas] Various terms for transport portion of a network slicing

 

Hi Lou, Pavan, Deborah and all,

 

As per TEAS WG chairs request during the last week’s meeting on network slicing, the co-authors of the definition draft had a few meeting and compiled all the potential terms with their pros and cons in the following table.

We feel that we captured all the terms suggested by TEAS and NSDT members in mailing lists and during the various meetings. Please let us know if any terms is missing.

 

In summary, the co-authors of the draft believe the term “IETF Network Slice” is a golden middle where IETF provides the exact context of any Network slice.

 

Cheers,

Reza (on behalf of all co-authors)

 




Various options for transport connectivity term from IETF point of view (without any specific order)


 

Term

Suggested by 

Pros

Cons


1

IETF Network Slice

TEAS chairs

o It clearly elaborates the scope of technologies addressed with in the IETF leveraging the industry-wide term 'network slice.
o It is golden middle, where “IETF” provides the exact context to “Network Slice”
o Acceptable to TEAS WG chairs
o Acceptable to draft co-authors

labeling a piece of work with SDO name is not a good idea and IETF has always worked towards wider use, generic solutions, so the name may be restrictive.


2

Transport Slice

Draft Authors and NSDT

o ‘Transport network’ is an abstraction of connectivity between the (network) end-points which is technology agnostic.  Well covered by RFC5921.

o Aligned with other SDO (i.e. MEF)
See Figure 17 of following white paper
https://wiki.mef.net/display/CESG/Slicing+for+Shared+5G+Fronthaul+and+Backhaul+-+White+Paper

As per recent WG adoption poll, it is not accepted


3

Transport network slice

Draft Authors and some IETF members

‘Transport network’ is an abstraction of connectivity between the (network)end-points which is technology agnostic.  Well covered by RFC5921.

As per recent WG adoption poll, it is not accepted


4

Connectivity Slice

Draft Authors

o Since the transport slice is a set of distinct connections, term "Connectivity Slice" is selected

o Aligned with other SDO (i.e. 3GPP) 
See Figure 4.9.3.1 of TR 28.801 and  http://www.3gpp.org/NEWS-EVENTS/3GPP-NEWS/1951-SA5_5G

As per TEAS WG chair, connectivity has different meaning at IETF


5

Connectivity Network Slice

Luis

The term Network becomes now narrow downed to the reference to connectivity, which is subject of IETF

As per TEAS WG chair, connectivity has different meaning at IETF


6

Connection Slice

Luis

o The term associates the concept of slice to the connection enabling the data transmission among end-points participating of a communication
o Note – here we could then follow a similar approach to how the VPNs are classified as L2 or L3; I mean L3/L2/(L1?) Connection Slice; if we classify the Connection Slices in that manner, such classification of the Connection Slice types will also help to describe recursiveness or hierarchical (multi-layer) slicing

o A connection can be established at different levels, including protocols above Layer 3
o Connection slice can make the people understand that there is a single connection represent a slice (i.e., 1:1) while actually could not be the case (i.e., 1 slice being formed by N connections)


7

Slice Network

Stewart 

Stewart is coming with some background that a slice is combination of storage, compute and communications (or network). Slice network means an existing network is sliced to serve a particular user-case.

According to Lou, it has entirely different meaning.


8

Virtual Slice Network (VSN)

Luis

o Variant on top of Stewart’s suggestion to link with the evolution of the concept of legacy VPNs
o The term reminds the idea of logical network per customer focusing on connectivity
o As before, it could be possible to follow a similar approach to how the VPNs are classified as L2 or L3; I mean L3/L2/(L1?) VSN; also here, such classification of the VSNs can also help to describe recursiveness or hierarchical (multi-layer) slicing in transport

o There is no specific reference to transport or connectivity, apart of the generic idea of network (which we now is also an overloaded term)
o Differently from a VPN, which basically is a single instance including a number of locations, a VSN could refer to a set of individual VSNs (e.g., one per network segment). So can be probably confusing. Thus probably it would be needed to add additional terms such as sub-VSN, or VSN-segment, VSN-part, VSN-sub-slice, or alike


9

TE Network slice 

Some TEAS members

Aligns with same  rationale used for naming ACTN.

Since not all transport networks are TE enabled, the realization of connectivity might be in a non-TE network. So, this term seems not appropriate


10

Carrier network slice

Webex/Stewart

In a generic use of term 'carrier' a carrier slice network carries use-case specific network traffic.

it may be confusing because it is associated with the infrastructure of telecommunication service providers, i.e., FNO/MNO. Recently, some network operators  deploy COTS servers in their infrastructure for MEC usages, and some readers may expect control of compute and storage  resources is in scope


11

Network slice

Some IETF members

An adoption of industry-wide term. While each SDO may look at it differently based on its own set of capabilities, for an end user it is a network slice in a specific technology domain.

o Since multiple connections are part of a single "Network Slice", it is not a good idea to call each of these connections "Network slice".
o There is a lack of 'harmonized' definition of network slice. For end customers, message may be confusing on which SDO they should ask for what part. It may lead to duplication of orchestration or APIs, depending upon who is controlling end to end network slice  - is it 3GPP operator, MVNO, ISP, service-integrator, OTT etc... 


12

Data transmission network slice (DTNS)

Shunsuke 

Since the transport slice is a set of distinct connections, providing the data transmission, this term might be suitable.  

 


13

Transmission Network  Slice

Reza

Since the transport slice provides the data transmission across transport network, this term might be suitable.  

 


14

Transmission Slice

Reza

Same as "Transmission Network Slice"