Re: [Teas] [Netslices] terminology discussion network slicing

[Gert Grammel <[e-mail]>]

Hi Daniele,

See below for more details:

On 2017-05-30, 10:17, "Daniele Ceccarelli" <[e-mail]> wrote:

    Ciao Gert,
    
    Please see in line
    Cheers
    Daniele  
    
    
    -----Original Message-----
    From: Teas [mailto:[e-mail]] On Behalf Of Gert Grammel
    Sent: lunedì 29 maggio 2017 15:34
    To: Daniele Ceccarelli <[e-mail]>; [e-mail]; 'Igor Bryskin' <[e-mail]>; 'Jeff Tantsura' <[e-mail]>; 'qiangli (D)' <[e-mail]>; 'Leeyoung' <[e-mail]>; [e-mail]
    Cc: [e-mail]
    Subject: Re: [Teas] [Netslices] terminology discussion network slicing
    
    Daniele,
    
    Trying to summarize what you're saying: a Network slice is an overlay network with minimum service guarantees, including virtual network functions available in this overlay network.  
    [>DC] why do you speak about an overlay network? If for example you have a site connected with a lambda to a data center and there you have a number of VNF, that's IMO a slice but there is no overlay. I'd say that there could be an overlay but that's not mandatory.
GGR> A Lambda travels through a WDM which is a Multiplexer (see discussion below). So a lambda is in my view an overlay, the multiplexed signal (OMS) being the underlay.    



    In your text I have issues with "dedicated to a given customer". Whether such construct is dedicated to one or shared among several legal customers shouldn't be of concern for the definition of a Network Slice.
    [>DC] Correct, it should be possible to have it dedicated or shared. We just need to make sure that both options are supported.  
GGR>> So if it can either be dedicated or shared, why mentioning this at all? Is there anything else?    

    On similar terms, whether a resource (being a NF or transport capacity) is shared or not is a design choice rather than a definition. In particular, Multiplexer by definition share some sort of common capacity irrespective whether it is TDM, Packet, WDM or CDM. So, the nature of multiplexing technique shouldn't be a definition criteria for a Network Slice either.
    
    Another food for thought:
    - I would also add that a network slice may use it's own internal addressing scheme isolated from other slices. While this is optional, I would expect it to be used quite often. 
    [>DC] agree. Ala VPN.

    - Why is it necessary to dedicate resources to a "Network Slice" at all? If e.g. a firewall NFV is in the DC and communicates over the Internet with a CPE (both belonging to a single owner) it would not fulfil the criteria of dedicated resource (Internet after all isn't dedicated), but it still looks pretty much like a "network slice".

    [>DC] Think of remote surgery...wouldn't you want every single part of the end to end network from the doctor's hands to the scalpel being totally dedicated? :) ...and in that case I don't think the packets would go through the internet
GGR>> There are a few issues here:
1)	this example looks a bit out-dated to me. There is only a one-to-one relationship between doctor and patient. This is what I would call a classical client-server model with protected resources. Ask for SONET leased line circuits if this is what a "slice" would mean. In defining what constitutes a slice you would also have to think about what constitutes a "doctor-slice" and what would be a "patient-slice"? (assuming responsible health care to avoid sarcastic interpretation)
2)	The question asked in my earlier email is whether the packets *must* go through a dedicated resource to call the thing "slice". In other words, if a thing that uses a non-dedicated "underlay" resource like the Internet can be called "slice" too? It is puzzling to me why the definition of "slice" would need any dedicated resource at all. It is in the end an implementers choice how much dedicated equipment is needed to provide a service. So why a slice has to use "dedicated resources", and what kind of "dedication" is meant? (See our earlier discussion about WDM)? Take your mobile as an example. It supports telephony and email. Those are different apps, potentially running in different VMs. 
a.	Is your phone a shared device or a dedicated one? 
b.	does using different apps means to use a different slices?
c.	If you write emails to different recipients, would that qualify your email to be "dedicated", hence a different slice, just because they are in different address books?
d.	If you switch from 3G to VoIP, does it mean to switch the slice, even if it is the same app?

GGR> My thinking is that the definition of "slice" needs to make practical sense. The 3GPP definition is probably a bit too high level but makes sense to me:
    "A set of network functions and the resources for these network functions which are arranged and configured, forming a complete logical network to meet certain network characteristics."

    Gert 
    
    On 2017-05-29, 11:01, "Teas on behalf of Daniele Ceccarelli" <[e-mail] on behalf of [e-mail]> wrote:
    
        Hi Sebastian, Jeff, Igor, all,
        
        I understand Jeff and Igor agree on the fact that a network slice does not necessarily include resources in the radio segment/domain and in the packet core one.
        An example could be an enterprise with dedicated resources in the "transport" segment (e.g. VPN, VN and TE-tunnels) that connect the customer premises to a data center where a number of VNF (e.g. firewall, nat, dpi) is dedicated to that given customer.
        My personal opinion is that this should be considered a network slice.
        
        The definition of network slice should allow for this scenario and mostly should clearly state whether this is included or not. 
        
        Sebastian, I don't understand if your statement: 
        " The management and orchestration of resources should be capable of instantiating a 'network slice' with an e2e (end-to-end) scope, where the scope could be between two devices, connected across intervening radio networks and core networks that may belong to the same or different administrative domains."
        implies that there must be a radio and a core segment in order to have a slice.
        
        BR
        Daniele  
        
        
        -----Original Message-----
        From: sebastian [mailto:[e-mail]] 
        Sent: domenica 28 maggio 2017 07:12
        To: 'Igor Bryskin' <[e-mail]>; 'Jeff Tantsura' <[e-mail]>; 'qiangli (D)' <[e-mail]>; Daniele Ceccarelli <[e-mail]>; 'Leeyoung' <[e-mail]>; [e-mail]
        Cc: [e-mail]
        Subject: RE: [Netslices] [Teas] terminology discussion network slicing
        
        Hello, Jeff,
        
        Agree with your view that the scope (core and/or radio layer) of a 'network slice' could vary depending on the triggers for instantiating the 'network slice'.
        
        On the other hand the instantiation of a 'network slice' may also be e2e (end-to-end), such as in device roaming scenarios, or in the case of vertical services that span different service providers.
        
        The span of a 'network slice' is only limited by the availability of resources and appropriate policies (e.g. SLAs), independent of whether those resources are in the core network layer or in the radio network layer.
        
        The management and orchestration of resources should be capable of instantiating a 'network slice' with an e2e (end-to-end) scope, where the scope could be between two devices, connected across intervening radio networks and core networks that may belong to the same or different administrative domains.
        
        Many thanks.
        -Sebastian
        
        
        -----Original Message-----
        From: Netslices [mailto:[e-mail]] On Behalf Of Igor Bryskin
        Sent: Wednesday, May 24, 2017 7:56 AM
        To: Jeff Tantsura <[e-mail]>; qiangli (D) <[e-mail]>; Daniele Ceccarelli <[e-mail]>; Leeyoung <[e-mail]>; [e-mail]
        Cc: [e-mail]
        Subject: Re: [Netslices] [Teas] terminology discussion network slicing
        
        Thanks Jeff,
        
        I am glad we are in sync.
        
        Igor
        
        -----Original Message-----
        From: Jeff Tantsura [mailto:[e-mail]] 
        Sent: Wednesday, May 24, 2017 8:45 AM
        To: Igor Bryskin; qiangli (D); Daniele Ceccarelli; Leeyoung; [e-mail]
        Cc: [e-mail]
        Subject: Re: [Teas] terminology discussion network slicing
        
        +1 Igor
        
        From transport side (L0-L3) we need to expose set of API’s that would allow to request a set of resources (a slice) either virtualized of physical that meet the constrains as requested by the business logic. Radio, while would definitely be one of most important use cases, would not be exclusively for radio use only. Any network management/ orchestration layer ideally should be able to use same set of API’s to request an e2e service.
        
        Cheers,
        Jeff
         
        
        On 5/24/17, 05:26, "Teas on behalf of Igor Bryskin" <[e-mail] on behalf of [e-mail]> wrote:
        
            Hi Cristina,
            
            I didn’t say that ns has nothing to do with radio. I said that ns *may* have nothing to do with radio, i.e. may support/enable network architectures that do not involve radio necessarily.
            
            Igor
            
            -----Original Message-----
            From: qiangli (D) 
            Sent: Wednesday, May 24, 2017 5:05 AM
            To: Igor Bryskin; Daniele Ceccarelli; Leeyoung; [e-mail]
            Cc: [e-mail]
            Subject: 答复: [Teas] terminology discussion network slicing
            
            Hi Igor,
            
            I share the same opinion on 5G and network slicing, but can't see the relation between this opinion and your conclusion that ns has nothing to do with radio.
            
            Best regards,
            
            Cristina QIANG
            
            
            -----邮件原件-----
            发件人: Igor Bryskin 
            发送时间: 2017年5月20日 23:50
            收件人: qiangli (D); Daniele Ceccarelli; Leeyoung; [e-mail]
            抄送: [e-mail]
            主题: RE: [Teas] terminology discussion network slicing
            
            Hi Cristina,
            
            People whose opinion I respect very much told me that even the 3GPP network slicing definition and effort is not 5G specific.  Network slicing is viewed as an important capability of 5G, but it could serve other architectures.
            
            Cheers,
            Igor
            
            -----Original Message-----
            From: qiangli (D)
            Sent: Saturday, May 20, 2017 5:27 AM
            To: Igor Bryskin; Daniele Ceccarelli; Leeyoung; [e-mail]
            Cc: [e-mail]
            Subject: 答复: [Teas] terminology discussion network slicing
            
            Hi Igor,
            
            I have different opinion on " Network slicing may have nothing to do with radio".
            IMO, network slicing should be a resource-oriented management mode. The end-to-end network slice (from UE to server/other UE) of course involves the radio resource, usually we use the word "RAN-radio access network" rather than simply say "radio". The Frequency band, time, code, equipment/site, and software are a number of dimensions, in which RAN resources can be isolated [Online available http://www.huawei.com/en/industry-insights/mbb-2020/trends-insights/5g-service-guaranteed-network-slicing-whitepaper].
            
            Best regards,
            
            Qiang, Li (Cristina)
            NetLab, Huawei
            
            
            -----邮件原件-----
            发件人: Netslices [mailto:[e-mail]] 代表 Igor Bryskin
            发送时间: 2017年5月20日 6:12
            收件人: Daniele Ceccarelli; Leeyoung; [e-mail]
            抄送: [e-mail]
            主题: Re: [Netslices] [Teas] terminology discussion network slicing
            
            Hi Danielle,
            
            Network slicing may have nothing to do with radio. 5g is just a big reason as to why the interest/focus is back on network slicing, which is a very old concept. I don't think that anyone is arguing with what you wrote. The questions are:
            a) how do you manage these "pieces"?
            b) what policy infrastructure should be put in place?
            c) how do you ensure their separation or, conversely, the degree of resource sharing?
            d) why exactly legacy VPNs are not good enough?
            
            Cheers,
            Igor
            
            
              
            
            -----Original Message-----
            From: Teas [mailto:[e-mail]] On Behalf Of Daniele Ceccarelli
            Sent: Friday, May 19, 2017 4:19 PM
            To: Leeyoung; [e-mail]
            Cc: [e-mail]
            Subject: Re: [Teas] terminology discussion network slicing
            
            Hi Young, all,
            
            i agree with your conclusion but would like to clarify one thing that IMO got lost in the discussion since its beginning.
            
            The 3GPP definition says: 
            "A set of network functions and the resources for these network functions which are arranged and configured, forming a complete logical network to meet certain network characteristics."
            
            This means that a network slice IS NOT a VPN or a TE Tunnel.
            A network slice is "something" (netslices and 3GPP will define what this something is) that is composed by a "piece" in the RADIO domain, a "piece" in the CLOUD domain, a "piece" in the TRANSPORT domain, plus possible other pieces in possible other domains.
            
            The word "transport" can be misleading here since one could think of transport technologies (e.g. WDM, OTN), but what I'm referring to as TRANSPORT DOMAIN is that part of the network that is used to carry a packet between two other domains. 
            In order to have a slice, that portion of the transport domain needs to be engineered, hence it is all about building a TE entity and stitching services to such entity. This is what is in the ACTN scope.
            
            My very personal opinion is that whatever belongs to the transport domain belongs to IETF (and is already being addressed), while the rest is a dangerous duplication of concepts standardized is other SDOs...but this is another discussion that doesn't fit here.
            
            BR
            Daniele  
            
            
            -----Original Message-----
            From: Teas [mailto:[e-mail]] On Behalf Of Leeyoung
            Sent: venerdì 19 maggio 2017 15:15
            To: [e-mail]
            Cc: [e-mail]
            Subject: Re: [Teas] terminology discussion network slicing
            
            Hi,
            
            Lou is right. There is a dedicated email list for the discussion of "network slicing (cc'ed)" and the discussion about what that term means should be held on that list. 
            
            We have used similar language in draft-ietf-teas-actn-framework right from the
            00 version. Recent changes have been an attempt to clarify what the terminology means in the context of ACTN. We are not trying to define or redefine "network slicing" in the ACTN document, but are trying to make it clear how ACTN works.
            
            Therefore I propose the following resolution:
            
            1. All discussion of the general applicability and definition of "network slicing" is held on the netslices mailing list.
            
            2. We adopt Adrian's suggestion to explain that the scope of the definition of the terms used in the ACTN framework is limited to ACTN. That means effectively that if there are components of a wider definition of network slicing that are not supported by ACTN, that will be OK. 
            
            I propose to post an updated version in the next few days and I believe that will allow this draft to move ahead without being blocked by the discussion in netslices. Once the IETF has a stable definition of network slicing we can return and see how ACTN is applicable to that definition an whether more wok is need (in a separate draft).
            
            Thanks,
            Young
            
            -----Original Message-----
            From: Lou Berger [mailto:[e-mail]]
            Sent: Tuesday, May 16, 2017 8:35 AM
            To: Igor Bryskin <[e-mail]>; Belotti, Sergio (Nokia - IT/Vimercate) <[e-mail]>; Gert Grammel <[e-mail]>; Leeyoung <[e-mail]>; [e-mail]; [e-mail]
            Subject: Re: [Teas] terminology discussion network slicing
            
            Perhaps it's time to bring the discussion to the slicing list and report back their reaponse....
            
            Lou
            
            
            On May 16, 2017 8:31:19 AM Igor Bryskin <[e-mail]> wrote:
            
            > Hi Sergio,
            >
            > I would also like to hear more from network slicing experts.
            >
            > My understanding is that the difference in the separation (in terms of 
            > control and data planes, security, etc.). For example, traditional BGP 
            > based L3 VPNs (that use provider's common control plane for their 
            > management and IP/MPLS TE tunnels to inter-connect their PEs) will 
            > probably not be able guarantee for the clients msec range connectivity 
            > setup times required by 5g, while provided by the same provider fully 
            > separated/genuinely private IP/MPLS networks (that do not share 
            > IP/MPLS control plane and infrastructure, whose network topology is 
            > supported by separate L0/L1 connections) hopefully will be able to 
            > provide such guarantees. Therefore, I define layer network slices as 
            > dynamically managed fully isolated in control and data planes private 
            > TE layer networks, which may share one or more underlying server layer networks.
            >
            >
            > Cheers,
            > Igor
            >
            >
            >
            >
            > -----Original Message-----
            > From: Teas [mailto:[e-mail]] On Behalf Of Belotti, Sergio 
            > (Nokia - IT/Vimercate)
            > Sent: Tuesday, May 16, 2017 6:08 AM
            > To: Gert Grammel; Leeyoung; [e-mail]; [e-mail]
            > Subject: Re: [Teas] terminology discussion network slicing
            >
            > Hi Gert,
            >
            > "Thinking a bit about it I came to the point where "VPN" and "network 
            > slices" seem to describe the same entity or at least a "network slice"
            > being a VPN of VPNs?"
            >
            > I share completely your conclusion , I'd like if someone can explain 
            > if a difference really exists .
            >
            > Thanks
            > Sergio
            >
            >
            > -----Original Message-----
            > From: Teas [mailto:[e-mail]] On Behalf Of Gert Grammel
            > Sent: Monday, May 15, 2017 7:02 PM
            > To: Leeyoung <[e-mail]>; [e-mail]; [e-mail]
            > Subject: Re: [Teas] terminology discussion network slicing
            >
            > Leeyoung,
            >
            > Thank you for taking a stab on this. Usually when getting to a 
            > definition, I try to establish what kind of existing constructs would 
            > fall under the definition. If my understanding is correct, the 
            > following list of constructs would all satisfy the definition somehow.
            > - A TDM network with a p2p TDM connection
            > - A PSC capable network carrying a p2p circuit (such as EPL/EVPL)
            > - An MPLS LSP using a traffic engineered IP network
            > - A L2VPN using a traffic engineered MPLS network
            > - A L3VPN using a traffic engineered IP network
            > - A TCP connection using a traffic engineered IP network
            > - Different QoS classes in an IP network
            >
            > Thinking a bit about it I came to the point where "VPN" and "network 
            > slices" seem to describe the same entity or at least a "network slice"
            > being a VPN of VPNs?
            >
            > Gert
            >
            >
            > On 2017-05-17, 16:44, "Teas on behalf of Leeyoung" 
            > <[e-mail] on behalf of [e-mail]> wrote:
            >
            >     Hi Adrian and others,
            >
            >     We'd like cross check with you on some terminology we introduced newly. Any 
            >     comment on these terms will be greatly appreciated.
            >
            >     We introduced 'network slicing' as follows:
            >
            >             Network slicing is a collection of resources
            >             that are used to establish logically dedicated virtual networks
            >             over TE networks. It allows a network provider to provide
            >             dedicated virtual networks for application/customer over a
            >             common network infrastructure. The logically dedicated
            >             resources are a part of the larger common network
            >             infrastructures that are shared among various network slice
            >             instances which are the end-to-end realization of network
            >             slicing, consisting of the combination of physically or
            >             logically dedicated resources.
            >
            >
            >     Thanks.
            >     Young and Daniele
            >     -----Original Message-----
            >     From: Leeyoung
            >     Sent: Friday, May 05, 2017 1:41 PM
            >     To: [e-mail]
            >     Subject: RE: [Teas] I-D Action: 
            > draft-ietf-teas-actn-framework-05.txt
            >
            >     Hi,
            >
            >     This update is intended to incorporate the comments from the last WG 
            >     meeting and any pending issues. We also have taken the global editorial 
            >     changes to make it consistent through the document. Major changes are:
            >
            >     - Inclusion of "network slicing" definition from ACTN perspective (in the 
            >     terminology section)
            >     - Added virtual network service (VNS) section (Section 3) to define types 
            >     of VNS.
            >     - Incorporated "orchestration" (service/network) mapping to ACTN 
            >     architecture (See Section 5.2)
            >     - Created a new section 6 (Topology Abstraction Method) where we imported 
            >     some texts from ACTN abstraction method  
            >     https://tools.ietf.org/html/draft-lee-teas-actn-abstraction-01
            >     - Added Appendices A & B to discuss example deployment scenarios such as 
            >     example of MDSC and PNC functions integrated in Service/Network 
            >     Orchestrator (Appendix A) and example of IP + Optical network with L3VPN 
            >     service (Appendix B)
            >
            >     In regard to ACTN abstraction method draft, we are going to keep it as a 
            >     separate draft and use this document to elaborate other aspects not 
            >     imported to the framework document.
            >
            >     The following diff pointer will help you see the changes with this revision:
            >     
            > https://www.ietf.org/rfcdiff?url2=draft-ietf-teas-actn-framework-05
            >
            >     The co-authors believe that the document is ready for WG LC. Any 
            >     changes/comments will be appreciated.
            >
            >     Thanks & Best regards,
            >     Young & Daniele (on behalf of other co-authors/contributors)
            >
            >     -----Original Message-----
            >     From: Teas [mailto:[e-mail]] On Behalf Of [e-mail]
            >     Sent: Friday, May 05, 2017 10:41 AM
            >     To: [e-mail]
            >     Cc: [e-mail]
            >     Subject: [Teas] I-D Action: draft-ietf-teas-actn-framework-05.txt
            >
            >
            >     A New Internet-Draft is available from the on-line Internet-Drafts directories.
            >     This draft is a work item of the Traffic Engineering Architecture and 
            >     Signaling of the IETF.
            >
            >             Title           : Framework for Abstraction and Control of Traffic 
            >             Engineered Networks
            >             Authors         : Daniele Ceccarelli
            >                               Young Lee
            >     	Filename        : draft-ietf-teas-actn-framework-05.txt
            >     	Pages           : 41
            >     	Date            : 2017-05-05
            >
            >     Abstract:
            >        Traffic Engineered networks have a variety of mechanisms to
            >        facilitate the separation of the data plane and control plane. They
            >        also have a range of management and provisioning protocols to
            >        configure and activate network resources.  These mechanisms
            >        represent key technologies for enabling flexible and dynamic
            >        networking.
            >
            >        Abstraction of network resources is a technique that can be applied
            >        to a single network domain or across multiple domains to create a
            >        single virtualized network that is under the control of a network
            >        operator or the customer of the operator that actually owns
            >        the network resources.
            >
            >        This document provides a framework for Abstraction and Control of
            >        Traffic Engineered Networks (ACTN).
            >
            >
            >
            >     The IETF datatracker status page for this draft is:
            >     https://datatracker.ietf.org/doc/draft-ietf-teas-actn-framework/
            >
            >     There are also htmlized versions available at:
            >     https://tools.ietf.org/html/draft-ietf-teas-actn-framework-05
            >     
            > https://datatracker.ietf.org/doc/html/draft-ietf-teas-actn-framework-0
            > 5
            >
            >     A diff from the previous version is available at:
            >     
            > https://www.ietf.org/rfcdiff?url2=draft-ietf-teas-actn-framework-05
            >
            >
            >     Please note that it may take a couple of minutes from the time of 
            >     submission until the htmlized version and diff are available at tools.ietf.org.
            >
            >     Internet-Drafts are also available by anonymous FTP at:
            >     ftp://ftp.ietf.org/internet-drafts/
            >
            >     _______________________________________________
            >     Teas mailing list
            >     [e-mail]
            >     https://www.ietf.org/mailman/listinfo/teas
            >
            >     _______________________________________________
            >     Teas mailing list
            >     [e-mail]
            >     https://www.ietf.org/mailman/listinfo/teas
            >
            >
            > _______________________________________________
            > Teas mailing list
            > [e-mail]
            > https://www.ietf.org/mailman/listinfo/teas
            >
            > _______________________________________________
            > Teas mailing list
            > [e-mail]
            > https://www.ietf.org/mailman/listinfo/teas
            >
            > _______________________________________________
            > Teas mailing list
            > [e-mail]
            > https://www.ietf.org/mailman/listinfo/teas
            >
            
            
            _______________________________________________
            Teas mailing list
            [e-mail]
            https://www.ietf.org/mailman/listinfo/teas
            
            _______________________________________________
            Teas mailing list
            [e-mail]
            https://www.ietf.org/mailman/listinfo/teas
            
            _______________________________________________
            Netslices mailing list
            [e-mail]
            https://www.ietf.org/mailman/listinfo/netslices
            _______________________________________________
            Teas mailing list
            [e-mail]
            https://www.ietf.org/mailman/listinfo/teas
            
        
        
        _______________________________________________
        Netslices mailing list
        [e-mail]
        https://www.ietf.org/mailman/listinfo/netslices
        
        _______________________________________________
        Teas mailing list
        [e-mail]
        https://www.ietf.org/mailman/listinfo/teas
        
    
    _______________________________________________
    Teas mailing list
    [e-mail]
    https://www.ietf.org/mailman/listinfo/teas