[Teas] Re: Last-gasp review of draft-ietf-teas-5g-ns-ip-mpls-11

[Adrian Farrel <[e-mail]>]

Hi Med,

 

Absent any follow-up by Monday, we will proceed with the publication of the
candidate version shared earlier.

 

Posting new versions is always a good idea.

 

A diff to track the changes made so far can be seen here:
https://author-tools.ietf.org/api/iddiff?doc_1=draft-ietf-teas-5g-ns-ip-mpls
<https://author-tools.ietf.org/api/iddiff?doc_1=draft-ietf-teas-5g-ns-ip-mpl
s&url_2=https://boucadair.github.io/5g-slice-realization/draft-ietf-teas-5g-
ns-ip-mpls.txt>
&url_2=https://boucadair.github.io/5g-slice-realization/draft-ietf-teas-5g-n
s-ip-mpls.txt.

 

[AF] Diff looks good. But two new nits.

 

3.4.2 now seems to have...

   This document focuses on RFC9543 9543 Network Slice deployments where

 

Just after Figure 16

s/(e.g., no LDP, RSVP, and SR)./ (e.g., no LDP, RSVP, or SR)./

 

Please see inline. 

 

Yes.

 

idnits shows " Unused Reference: 'TR-GSTR-TN5G' "

 

[Med] Argh. Fixed. Thanks.

 

---

 

Section 2, having said the document makes use of the terms defined in

RFC 9543, proceeds to provide a new definition of "customer" and 

"provider". I wonder whether those new definitions are necessary, and 

if they are I think great care is needed to achieve disambiguation.

 

[Med] Section 2 also says right after the ref to 9543: "See Section 3.3 for
the contextualization of some of these terms."

 

We used to have "customer" and "provider" entries be part of 3.3 but we
moved them here because we received a comment to group all defs.

 

[AF] OK. *If* you must have definitions of customer and provider that
differs from what is in 9543, then including the definitions in Section 2 is
the right thing to do. I think, in that case, the forward reference to 3.3
is fine.

 

But, your definition of customer *is* different from the definition in 9543.
You call out that this is different from a 5G customer by saying "This
entity is distinct from the customer of a 5G Network Slice Service." But you
don't note the difference from 9543.

 

I am unable to tell whether this is a scoping difference (yours is a subset
of the 9543 definition), or something different. I don't really find 3.3
clarifies this, and I think that, given the statement that this document
uses the terms defined in 9543, you need some explanation. For example, you
might say "In the context of this document, this definition replaces the one
found in RFC 9543," or whatever is appropriate.

 

The definition of provider seems to be much closer to that in 9543. Close
enough that I wonder why you need is.

 

In Section 3.1, piping ("|") is used to indicated quoted text from 3GGPP

documents. That's good.

But: 

- It is probably de trop to use quotation marks as well.

[Med] "" were used to ease demux the quotation from the source.

- It appears as though the reference is part of the quote.

 

[Med] moved the source refs to be positioned before the citation and removed
the redundant "".

---

 

3.2.1

 

Ditto the use of | and "

[Med] Fixed.

 

---

3.2.2

 

I don't object to one word of the text in 3.2.2. I do find it peculiar

that, while 3.2.1 is nicely quoted from 3GPP documentation, the text in

this section is longer and un-cited.

 

I should imagine that the paragraph...

    The objective of Transport Network Slicing is

...should be founded in a 3GPP definition.

 

[Med] We can't formally cite any reference here where the exact words are
used. The definition inspires from 3GPP docs, especially from the following:


 

"When providing an end to end communication service, the network may use
non-3GPP parts (e.g. Data centre network (DCN), Transport network (TN)) in
addition to the network components defined in 3GPP. Therefore, in order to
ensure the performance of a communication service according to the business
requirements, the 3GPP management system has to coordinate with the
management systems of the non-3GPP parts (e.g., MANO system) when preparing
a network slice for this service. This coordination may include obtaining
capabilities of the non-3GPP parts and providing the slice specific
requirements and other requirements on the non-3GPP parts. Figure 4.7.1
illustrates an example for the coordination with management of TN part
(e.g., directly or via MANO system)."

 

+ (extract of TN specific objectives)

 

"The network slice related requirements include requirements such as: area
traffic capacity, charging, coverage area, degree of isolation, end-to-end
latency, mobility, overall user density, priority, service availability,
service reliability, UE speed."

 

While the first half of the paragraph...

   Transport Network Slicing provides various degrees of sharing of

   resources between slices

...presumably stems from 9543

 

[Med] We can cite Section 8 of 9543.

 

[AF] Nice

 

The rest of the text is explanatory in the context of this document (and

maybe should say so).

 

[Med] Added the following NEW:

 

"The following further elaborates on how Transport Network Slicing is

defined in the context of this document."

 

[AF] That's helpful.

 

I wonder whether, in view of your statement that, "The definition inspires
from 3GPP docs," you might include some such statement with references to
the two documents you quoted from. Thus...

 

   The following further elaborates on how Transport Network Slicing is

   defined in the context of this document. It draws on the 3GPP definition

   of Transport Network Slicing as described in [ref] and [ref].

 

---

 

Slice scopes and SDP locations.

 

It is all really confusing to me, and I do not arrive at anything other

than contradictions from the text :-(

 

Figure 1 shows the TN slice and also shows the type 3 and type 4 SDP

locations.

[Med] These SDPs for "RFC 9543 Network Slice", not the full TN slice. We
have the following in the introduction:

 

   Specifically, this document describes an approach to

   how RFC 9543 Network Slices are realized within provider networks and

   how such slices are stitched to Transport Network resources in a

   customer site in the context of Transport Network Slices (Figure 1).

   Concretely, the realization of an RFC 9543 Network Slice (i.e.,

   connectivity with performance commitments) involves the provider

   network and partially the AC (the PE-side of the AC).  This document

   assumes that the customer site infrastructure is over-provisioned and

   involves short distances (low latency) where basic QoS/scheduling

   logic is sufficient to comply with the Service Level Objectives

   (SLOs).

 

But, an SDP is the point at which the slice service begins or ends. It

is the edge of the slice.

 

Either:

- You are saying that a transport slice does not map perfectly to an

  IETF slice

Or:

- There is something wrong with this and the text in 3.4.2.

 

I don't think any of this changes any of the technical details or

practicalities of the solutions described. It just that the description

seems plain wrong.

 

For example, where you talk about mapping service parameters: where

is that mapping performed? The question, I suppose is, are you 

constructing a TN slice that encompasses network resources that are

both IETF and non-IETF. In that case, the service parameters are

mapped at the edge of the TN slice, and then they are mapped again

at the network boundary at the start of the IETF technology.

 

If your intention is that the TN slice is constructed from elements of

the customer domain (possibly non-IETF) and the provider domain (IETF),

that would be fine, but I think the TN slice end points are still the 

SDPs.

 

If, conversely, you are saying that the TN slice is a composite of

customer domain slice and IETF slice, then you have the SDPs correctly

placed, but you have not made this composite clear.

 

[Med] 3.4.1 says: 

 

   A TN slice relies upon resources that can involve both the provider

   and customer TN domains.  More details are provided in Section 3.4.2.

 

   A TN slice might be considered as a variant of horizontal composition

    of Network Slices mentioned in Appendix A.6 of [RFC9543].

 

[AF] Yeah OK. And I see para 2 of the Introduction does say this. Maybe it
is clear enough. 

I'm not sure why it escaped me on first (or even nth) reading.

 

Maybe the only thing I am left with is to be careful all the way through
that when you talk about a network slice, we need to be clear whether you
are talking about the end-to-end (composed) TN slice, or just the 9543
slice. Perhaps worth a quick re-read to see whether there remains any
ambiguity.

 

[Med] Updated 3.4.2 to remind that this is about 9543 network slice
deployments.

 

It does not help that 3.4.2 goes on to say:

   The concept of distributed PE (Section 3.3.4) assimilates CE-based

   SDPs defined in Section 5.2 of [RFC9543] (i.e., Types 1 and 2) as SDP

   Type 3 or 4 in this document.

 

The problem is not with the text, but with the fact that your Figure 6

shows the TN slice extending to the CEs in a model where the CEs are

distinct from the PEs. (To me it seems that you are doing types 1 and 2

in this document with the case of distributed PE assimilating types 3

and 4.)

 

I start to run into some of this again in Section 4. For example, in 4.1

   In this option, the RFC 9543 Network Slice, fulfilling connectivity

   requirements between NFs that belong to a 5G slice, is represented at

   an SDP by a VLAN ID (or double VLAN IDs, commonly known as QinQ), as

   depicted in Figure 12.

That suggests to me that the model in your head is that the VLAN ID is 

indicative of either:

- onto which IETF slice the SDP should place the traffic (the traffic 

  is at that stage not yet sliced only marked for slicing)

or:

- on which IETF slice the traffic is flowing (the traffic has already

  been sliced and the SDP is actually at the end of the TN slice).

 

To complete my whining about this, 4.2 shows the "tunnels representing

slices" as extending all the way to the NFs (CEs not marked). Surely,

in that case, the SDP is at the end of the tunnel?

 

I'd be quite happy with either or any approach. I just want the 

description to match.

 

---

 

3.6

 

   At the time of writing (2024), Section 6.1.2 of [NG.113] specifies

 

Well...

NG.113 is a dated reference. Doesn't that mean that you can simply say

 

   Section 6.1.2 of [NG.113] specifies

 

While it is true that 3GGP's "5GS Roaming Guidelines" may change over

time. The referenced Version 4 dated 2021 will not change.

 

[Med] Fixed. 

---

 

3.7

 

s/of [RFC9522])./of [RFC9522]./

[Med] Fixed. Thanks.

 

---

 

Figure 11

 

Is there an arrow head missing from the left hand end of the line within

the right hand P node?

[Med] Thanks for catching this. There was a missing +.

 

[AF] Yeah, you caught something I missed (in the right hand PE a missing +),
but have not fixed the thing I mentioned (in the right hand P missing a <).

The line should read

== == |    +---->o<----->o<--->o<------>o<--->o<----->o<----+    |