Re: [Teas] network Slice Endpoint in draft-ietf-teas-ietf-network-slice-definition-00

[Adrian Farrel <[e-mail]>]

Hi Tarek,

 

[External Email. Be cautious of content]

 

[AF] I am always cautious of content, but never more so than when it comes with a warning attached.

 

1.       I think John’s text is a very good starting point. Thanks for that. Stick it in the draft and we can polish it.

2.       Connection constructs are not simply rooted at *a* CE because mp2p, mp2mp.

[TS]: I’m quoting from below, to realize the MP2P, one would require N-1 P2P unidirectional connections (on sending CEs), and 1 P2MP unidirectional connection on the hub.

“    If an IETF network slice service customer wants to have hub and spoke connectivity between N CEs in order to control how traffic is distributed between its CEs, it requests a set of N - 1 P2P unidirectional connectivity matrices, each between a sending CE spoke and the hub CE, and a P2MP connectivity matrix between the sending CE hub and the spoke CEs.“ 

[AF] Well, yes, that’s a quote, but it’s a quote about hub and spoke, not about mp2pm in general. He only reason you’d do that is if you want to control/specify a hub CE. Elsewise, you’d just ask for an MP2MP connectivity service and let the network get on with it.

3.       Yes, a slice service may include multiple connection constructs

[TS]: OK, good.

 

4.       Yes, an SLO applies to a sender’s data within a single connection construct. A connection construct may have one sender (p2p, p2mp) or multiple senders (mp2p, mp2mp).

[TS]: with the two basic constructs (unidirectional connection, and connection group) rooted on the respective CE, one can realize the above connectivity..

 

[AF] Realizing connectivity is, of course, interesting both to the vendor and the service provider, but it is a very different thing from the requested service. Yes, you could ask a customer to work out that they need n p2p connections and how to group them to build the service they want, but you wouldn’t (if you wanted their business). You would ask them to specify the set of senders and the set of receivers, and you (the operator) would work out what to do under the covers.

 

[AF] I have some sympathy with noting that p2p is only a special case of mp2mp, and that leads me to note that the general case can be stated as a list of senders and a list of receivers (with overlap of the sets allowed) and no need to describe the connectivity construct. On the other hand, people generally like the short hand that a connectivity construct yields.

 

5.       Yes, a bidirectional connection construct may exist, but it doesn’t have to have symmetric SLOs. (Actually it is an mp2mp services with only two end points.) So, yes, two SLOs for a bidirectional p2p connections, and those SLOs may be identical.

[TS]: Agreed, but to realize bidirectional connectivity, I require 2 unidirectional connection constructs whose SLOs may or may not be same.

 

[AF] I believe that is what I wrote, yes.

 

Cheers,

Adrian

 

From: Teas <[e-mail] <mailto:[e-mail]> > On Behalf Of Greg Mirsky
Sent: 04 April 2021 04:04
To: Tarek Saad <tsaad=[e-mail] <mailto:tsaad=[e-mail]> >
Cc: Kireeti Kompella <[e-mail] <mailto:[e-mail]> >; Manish Gupta <[e-mail] <mailto:[e-mail]> >; [e-mail] <mailto:[e-mail]> ; John E Drake <jdrake=[e-mail] <mailto:jdrake=[e-mail]> >; Joel M. Halpern <[e-mail] <mailto:[e-mail]> >; [e-mail] <mailto:[e-mail]> 
Subject: Re: [Teas] network Slice Endpoint in draft-ietf-teas-ietf-network-slice-definition-00

 

Hi Tarek,

thank you for bringing up these questions. I've been thinking, playing in my mind with the SLO in a network slice.

I agree that the SLO is for an ordered pair of CEs, and thus it is unidirectional regardless of the connection matrix. But I think that there might be a use case for a bidirectional P2P connection, especially if [A, B] and [B, A] SLOs are identical. Also, a bidirectional P2P connection might be a useful construct when directions are sharing each other fate under the protection switchover scenario.

 

What are your thoughts?

 

Regards,

Greg

 

On Sat, Apr 3, 2021 at 7:31 PM Tarek Saad <tsaad=[e-mail] <mailto:[e-mail]> > wrote:

John/all,

I can glean the following from this definition. Let me know if misinterpreted.

1) A connection supporting a network slice is unidirectional. A corollary is: SLO(s) are unidirectional too.

2) >From a sending CE, there are only 2 types of connectivity matrices:

*     P2P unidirectional

*     P2MP unidirectional

Using the above type of matrices on a sending CE, one can realize any of network slice connectivity type required – e.g., quoting your example, a MP2P can be realized by (N – 1) P2P unidirectional connectivity matrices on the sending CEs.

 3) The SLOs are per matrix type – either P2P or P2MP. Is the definition anywhere precluding having multiple connectivity types of matrices from a sending CE for the same network slice?(e.g. a mix of (n x P2P) and m x P2MP connectivity matrices on the same sending CE for the same network slice)?

 Regards,

Tarek

 On 4/3/21, 4:06 PM, "Teas on behalf of John E Drake" <[e-mail] <mailto:[e-mail]>  on behalf of jdrake=[e-mail] <mailto:[e-mail]> > wrote:

     Hi,

     As a follow-up to the email, below, that Eric and I sent in late February, here is our proposed definition of an IETF Network Slice Service:

     IETF Network Slice Service - A service provider instantiates a service for a customer which is specified in terms of a set of the customer's endpoints (CEs), a set of connectivity matrices (MP2MP, P2MP, MP2P, and P2P) between subsets of these CEs and an SLO for each CE sending to each connectivity matrix.  I.e., 

in a given IETF network slice service there may be multiple connectivity matrices of the same or different type, each connectivity matrix may be between a different subset of CEs, and for a given connectivity matrix each sending CE has its own SLO and each SLO may be different.  It is also the case that a given sending CE may have a different SLO for each connectivity matrix to which it is sending.  Note that a given sending CEs's SLO for a given connectivity matrix applies between it and each of the receiving CEs for that connectivity matrix. 

     This results in the following connectivity matrices:

              For a MP2MP connectivity matrix with N CEs, each of the N sending CEs has its own SLO and each may be different

              For a P2MP connectivity matrix, there is only one sending CE and there is only one SLO

              For a MP2P connectivity matrix with N CEs, each of the N - 1 sending CEs has its own SLO and each may be different

              For a P2P unidirectional connectivity matrix, there is only one sending CE and there is only one SLO

              For a P2P bidirectional connectivity matrix, there are two sending CEs, there are two SLOs, and each may be different 

    If an IETF network slice service customer wants to have hub and spoke connectivity between N CEs in order to control how traffic is distributed between its CEs, it requests a set of N - 1 P2P unidirectional connectivity matrices, each between a sending CE spoke and the hub CE, and a P2MP connectivity matrix between the sending CE hub and the spoke CEs. 

     It should be noted that per [RFC4364} section 9 (https://datatracker.ietf.org/doc/html/rfc4364#section-9 <https://urldefense.com/v3/__https:/datatracker.ietf.org/doc/html/rfc4364*section-9__;Iw!!NEt6yMaO-gk!WHFr7oDhtlR3DK2AParJ1Ca-QP2Y0pbvYA1LAeSQO1_0Fa29p8kgaurw0vnG2Q$> ) an IETF network slice service customer may actually be its own IETF network slice service provider in the same or different provider network.

     For a given IETF network slice service, the IETF network slice customer and the IETF network slice provider agree, on a per-CE basis which end of the attachment circuit provides the service demarcation point.  This determines whether the attachment circuit is included in any SLOs for the subject CE.

 

Juniper Business Use Only