Re: [Teas] FW: New Version Notification for draft-bestbar-teas-ns-packet-01.txt

[Igor Bryskin <[e-mail]>]

Pavan,

I assume:
- that a node, generally speaking, needs to make a forwarding decision based on the packet's slice topology;
- said topology may change dynamically, and so is the slice membership in the slice aggregate.
Therefore. It is better, in my opinion, to carry slice ID in packets and let nodes (re-)define slice aggregate membership independently.

Igor

Get Outlook for Android<https://aka.ms/ghei36>

________________________________
From: Teas <[e-mail]> on behalf of Vishnu Pavan Beeram <[e-mail]>
Sent: Sunday, February 7, 2021, 11:48 AM
To: Igor Bryskin
Cc: Tarek Saad; Gyan Mishra; TEAS WG; [e-mail]
Subject: Re: [Teas] FW: New Version Notification for draft-bestbar-teas-ns-packet-01.txt

Igor, Hi!

Please see inline.

Regards,
-Pavan (as a co-author)

On Sun, Feb 7, 2021 at 8:03 AM Igor Bryskin <[e-mail]<mailto:[e-mail]>> wrote:
Gyan and Pavan,

When a packet arrives at a node participating in a multi-slice network, the packet needs to be treated and *routed* according to the slice it belongs to. Because the physical data link over which the packet has arrived could be used by more than one slices, the packet needs to carry somewhere the slice ID. IMHO because the slice topology may dynamically change, it would be simpler and cleaner to carry the slice ID, rather than slice aggregate ID, as suggested by the draft.

[VPB] The packet needs to carry some identifier that would help determine the specific forwarding treatment (scheduling and drop policy) to be applied before the packet is forwarded further (this is not needed in the "Control Plane Slice Policy" mode). In the model that is proposed in the draft, we have one or more IETF Network Slices mapped onto a single slice aggregate and each slice aggregate has a specific Per Hop Behavior associated with it. In other words, the packet needs to carry an identifier that maps onto the slice aggregate. Also note that in the proposed model, multiple slice aggregates can map onto the same topology.


I also agree with Gyan that the COS field may not be the best choice for the task.

[VPB] I'm not sure what you mean by the COS field here.  If there is a need to allow differentiation of forwarding treatments for packets within a slice aggregate, the slice aggregate traffic may further carry a Diffserv Class Selector.


Regards,
Igor

Get Outlook for Android<https://aka.ms/ghei36>
________________________________
From: Teas <[e-mail]<mailto:[e-mail]>> on behalf of Vishnu Pavan Beeram <[e-mail]<mailto:[e-mail]>>
Sent: Thursday, February 4, 2021 3:49:37 PM
To: Gyan Mishra <[e-mail]<mailto:[e-mail]>>
Cc: Tarek Saad <tsaad=[e-mail]<mailto:[e-mail]>>; [e-mail]<mailto:[e-mail]> <[e-mail]<mailto:[e-mail]>>; TEAS WG <[e-mail]<mailto:[e-mail]>>
Subject: Re: [Teas] FW: New Version Notification for draft-bestbar-teas-ns-packet-01.txt

Gyan, Hi!

Thanks for taking the time to review the document. Please see inline for responses (prefixed VPB).

Regards,
-Pavan (as a co-author)

On Wed, Feb 3, 2021 at 8:14 PM Gyan Mishra <[e-mail]<mailto:[e-mail]>> wrote:


Dear authors

I reviewed the draft and have some comments.

I noticed the draft  name and throughout the document it mentions Network Slice in IP/MPLS networks and wonder if it would be more appropriate to say network slice in SR/MPLS networks.  As NS involves traffic steering either RSVP-TE or SR steering my thoughts are it would be appropriate

[VPB] I’m not sure if I understood the comment. As you are aware, Segment Routing can be instantiated on MPLS or IPv6 dataplanes. The solution proposed in draft-bestbar-teas-ns-packet is intended for realizing network slicing in IP(v4/v6) / MPLS networks. The proposed solution works with any type of path control technology (e.g.RSVP-TE paths, SR paths, FlexAlgo paths). If you are interested in how the proposed solution works in SR deployments, please refer to draft-bestbar-spring-scalable-ns.


This drafts main focus it seems is an example of NS using QOS CS selector marking Differential service to be used for network slicing.


[VPB] Not really. The focus in this document is on the Slice Policy construct which includes rules that control the following slice aggregate attributes:
- Data plane policies (Slice Selectors, QOS profiles)
- Control plane policies (Guaranteed BW, Reservation priorities)
- Topology membership policies (Customized/Pre-defined)

There are 3 types of Slice Policy Modes depending on how/where the shared network resources are partitioned –
(1) Data Plane Slice Policy Mode
(2) Control Plane Slice Policy Mode
(3) Data and Control Plane Slice Policy Mode

In modes (1) and (3), a Slice Selector (SS) is carried in the packet to identify the slice aggregate and apply specific forwarding treatment (scheduling treatment and drop probability). But this is not needed in mode (2).

In my mind QOS is completely orthogonal to the concept of network slicing.  QOS involves shared resources and traffic classification and scheduling based on shared pipe resources which is completely orthogonal to network slicing which is a cross section of underlying resources involving a degree of isolation during provisioning to meet a customer  SLA.  QOS has been around for a long time and the big downside to QOS is that it is independent of underlying network resources.

When QOS PHB scheduling occurs based on CS AF or EF selector, packet are marked at the edge of trust boundary and PHB hop by hop scheduled matching dscp value providing bandwidth guarantees for profile traffic at or below the scheduling bandwidth which is based on the shared physical link.



[VPB] As noted above, it is certainly possible for a network slicing solution to be put together without requiring a specific per-hop-behavior to be applied on packets belonging to a slice aggregate (we refer to this as “Control Plane Slice Policy Mode” - Section 4.2 of draft-bestbar-teas-ns-packet-01). However, there are cases where the Slice service requires strict QOS guarantees and differentiation from other traffic – this is where the Data Plane Slice Policy Mode comes into play.

Kind Regards

Gyan


On Tue, Dec 22, 2020 at 1:59 PM Tarek Saad <tsaad=[e-mail]<mailto:[e-mail]>> wrote:

Hi WG,



We have published a new revision of this draft (just in time for your holidays reading).

The changes include:

  *   additional co-authors have joined
  *   introduction of new terms “slice policy” and “slice aggregate”
  *   editorial nits to align with new terms introduced



As usual, reviews and comments are welcome.



Regards,

Tarek (on behalf of co-authors)





On 12/22/20, 1:42 PM, "[e-mail]<mailto:[e-mail]>" <[e-mail]<mailto:[e-mail]>> wrote:



    [External Email. Be cautious of content]





    A new version of I-D, draft-bestbar-teas-ns-packet-01.txt

    has been successfully submitted by Tarek Saad and posted to the

    IETF repository.



    Name:           draft-bestbar-teas-ns-packet

    Revision:       01

    Title:          Realizing Network Slices in IP/MPLS Networks

    Document date:  2020-12-22

    Group:          Individual Submission

    Pages:          27

    URL:            https://urldefense.com/v3/__https://www.ietf.org/archive/id/draft-bestbar-teas-ns-packet-01.txt__;!!NEt6yMaO-gk!UopixISsH-6laFtUNIg5vIaLsXzFthS8C65TCDCRk-5ayO6AAa-8IPzvuhQQRw$

    Status:         https://urldefense.com/v3/__https://datatracker.ietf.org/doc/draft-bestbar-teas-ns-packet/__;!!NEt6yMaO-gk!UopixISsH-6laFtUNIg5vIaLsXzFthS8C65TCDCRk-5ayO6AAa-8IPwWnIEvoQ$

    Htmlized:       https://urldefense.com/v3/__https://datatracker.ietf.org/doc/html/draft-bestbar-teas-ns-packet__;!!NEt6yMaO-gk!UopixISsH-6laFtUNIg5vIaLsXzFthS8C65TCDCRk-5ayO6AAa-8IPxBddbTdg$

    Htmlized:       https://urldefense.com/v3/__https://tools.ietf.org/html/draft-bestbar-teas-ns-packet-01__;!!NEt6yMaO-gk!UopixISsH-6laFtUNIg5vIaLsXzFthS8C65TCDCRk-5ayO6AAa-8IPznNKoLDg$

    Diff:           https://urldefense.com/v3/__https://www.ietf.org/rfcdiff?url2=draft-bestbar-teas-ns-packet-01__;!!NEt6yMaO-gk!UopixISsH-6laFtUNIg5vIaLsXzFthS8C65TCDCRk-5ayO6AAa-8IPzsN9oeaw$



    Abstract:

       Network slicing provides the ability to partition a physical network

       into multiple logical networks of varying sizes, structures, and

       functions so that each slice can be dedicated to specific services or

       customers.  Network slices need to operate in parallel while

       providing slice elasticity in terms of network resource allocation.

       The Differentiated Service (Diffserv) model allows for carrying

       multiple services on top of a single physical network by relying on

       compliant nodes to apply specific forwarding treatment (scheduling

       and drop policy) on to packets that carry the respective Diffserv

       code point.  This document proposes a solution based on the Diffserv

       model to realize network slicing in IP/MPLS networks.









    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<http://tools.ietf.org>.



    The IETF Secretariat






Juniper Business Use Only

_______________________________________________
Teas mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/teas
--

[http://ss7.vzw.com/is/image/VerizonWireless/vz-logo-email]<http://www.verizon.com/>

Gyan Mishra

Network Solutions Architect

M 301 502-1347
13101 Columbia Pike
Silver Spring, MD

_______________________________________________
Teas mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/teas