Re: [Detnet] Transport sub-layer name change (Was Re: Tsvart last call review of draft-ietf-detnet-architecture-08)

[Lou Berger <[e-mail]>]

IPv6 *is* there ;-)

On 12/13/2018 3:29 PM, Andrew G. Malis wrote:
> Lou,
>
> Thanks, that's what I thought, but I didn't see it in your list of 
> DetNet Forwarding sub-layer protocols.
>
> Cheers.
> Andy
>
>
> On Thu, Dec 13, 2018 at 1:55 PM Lou Berger <[e-mail] 
> <mailto:[e-mail]>> wrote:
>
>     Hi Andy,
>
>     As SRv6 is implemented using an IPv6 routing header the figure below
>     accurately shows IPv6 (and consequently SRv6) as being  part of the
>     Forwarding sub-layer.
>
>     In case you were really asking if use of SRv6 is in  scope of
>     DetNet WG,
>     I don't see anything in the charter that would preclude the WG
>     working
>     on DetNet SRv6.
>
>     Lou
>
>     On 12/13/2018 1:16 PM, Andrew G. Malis wrote:
>     > Lou,
>     >
>     > I think we've discussed this before, but I wanted to confirm
>     whether
>     > SRv6 is in scope for the DetNet Forwarding sub-layer.
>     >
>     > Thanks,
>     > Andy
>     >
>     >
>     > On Thu, Dec 13, 2018 at 12:47 PM Lou Berger <[e-mail]
>     <mailto:[e-mail]>
>     > <mailto:[e-mail] <mailto:[e-mail]>>> wrote:
>     >
>     >     Hi,
>     >
>     >          My reading is that the WG has settled on:
>     >
>     >            +----------------------------+
>     >            |  DetNet Service sub-layer  | PW, UDP, GRE
>     >            +----------------------------+
>     >            | DetNet Forwarding sub-layer| IPv6, IPv4, MPLS TE LSPs,
>     >     MPLS SR
>     >            +----------------------------+
>     >                          .
>     >                          .
>     >
>     >                        Figure 4: DetNet adaptation to data plane
>     >
>     >     Authors, (of detnet-architecture)
>     >          Please work to have the next rev of the draft reflect this
>     >     change.
>     >
>     >     Thank you to all who participated in the discussion!
>     >     Lou
>     >     (as co-chair and doc shepherd)
>     >
>     >     On 11/20/2018 1:18 PM, Lou Berger wrote:
>     >     > ALL,
>     >     >
>     >     > There is a desire to replace the word "Transport" from the
>     DetNet
>     >     > Transport sub-layer to avoid confusion with L$ Transport
>     protocols.
>     >     >
>     >     > In the TEAS WG we had a similar discussion and we replaced
>     >     "Transport"
>     >     > with "Traffic Engineered (TE) ".
>     >     >
>     >     > While a bit more verbose, what do people think about this
>     change?
>     >     >
>     >     > To be clear, the suggestion is:
>     >     >
>     >     > OLD
>     >     >
>     >     >                      .
>     >     >                      .
>     >     >        +----------------------------+
>     >     >        |  DetNet Service sub-layer  | PW, UDP, GRE
>     >     >        +----------------------------+
>     >     >        | DetNet Transport sub-layer | IPv6, IPv4, MPLS TE
>     LSPs,
>     >     MPLS SR
>     >     >        +----------------------------+
>     >     >                      .
>     >     >                      .
>     >     >
>     >     >                    Figure 4: DetNet adaptation to data plane
>     >     >
>     >     > NEW
>     >     >
>     >     >                      .
>     >     >                      .
>     >     >        +----------------------------+
>     >     >        |  DetNet Service sub-layer  | PW, UDP, GRE
>     >     >        +----------------------------+
>     >     >        |      DetNet TE sub-layer   | IPv6, IPv4, MPLS TE
>     LSPs,
>     >     MPLS SR
>     >     >        +----------------------------+
>     >     >                      .
>     >     >                      .
>     >     >
>     >     >                    Figure 4: DetNet adaptation to data plane
>     >     >
>     >     > Lou
>     >     >
>     >     > On 11/20/2018 11:21 AM, Pascal Thubert (pthubert) wrote:
>     >     >> Hello Lou:
>     >     >>
>     >     >>
>     >     >> About ' discuss changing the name of the "DetNet Transport
>     >     sub-layer"  to avoid the word "transport".  '
>     >     >>
>     >     >> For one I'd like to make that call. The unfortunate name
>     >     collision has started to hurt us quite a bit already and people
>     >     are getting confused on very active exchanges we have on the
>     >     mailing list.
>     >     >>
>     >     >> I tend to agree that for the general IETF "transport"
>     generally
>     >     means "L4". Even point one in your email uses "transport"
>     that way
>     >     I guess. Sadly many alternate names are highly overloaded
>     already
>     >     (think "carrier" for instance, or "bus"). I like the term
>     "train"
>     >     because of the association with a schedule, but that's just me.
>     >     >>
>     >     >> Same goes actually for the complex path that we build. That
>     >     complex path can be an elongated DODAG with multiple PREOF
>     points.
>     >     Usual terms like "circuit" or "path" fail to capture that
>     >     complexity. 6TiSCH found the term "Track" to refer to it.
>     >     >>
>     >     >> Would you push that discussion to the ML?
>     >     >>
>     >     >> Take care,
>     >     >>
>     >     >> Pascal
>     >     >>
>     >     >>> -----Original Message-----
>     >     >>> From: Lou Berger <[e-mail]
>     <mailto:[e-mail]> <mailto:[e-mail]
>     <mailto:[e-mail]>>>
>     >     >>> Sent: mardi 20 novembre 2018 13:11
>     >     >>> To: Scharf, Michael <[e-mail]
>     <mailto:[e-mail]>
>     >     <mailto:[e-mail]
>     <mailto:[e-mail]>>>
>     >     >>> Cc: [e-mail] <mailto:[e-mail]>
>     <mailto:[e-mail] <mailto:[e-mail]>>;
>     > [e-mail] <mailto:[e-mail]> <mailto:[e-mail]
>     <mailto:[e-mail]>>; draft-ietf-detnet-
>     >     >>> [e-mail]
>     <mailto:[e-mail]>
>     <mailto:[e-mail] <mailto:[e-mail]>>
>     >     >>> Subject: Re: [Detnet] Tsvart last call review of
>     >     draft-ietf-detnet-architecture-08
>     >     >>>
>     >     >>> Michael,
>     >     >>>
>     >     >>> I think we're getting somewhere and identifying where we
>     have
>     >     disconnects
>     >     >>> and what may (and what may not) need to change in the
>     >     document.  My
>     >     >>> takeaways are:
>     >     >>>
>     >     >>> - The document needs a good 'scrub' of the congestion
>     related
>     >     references to
>     >     >>> ensure that the document only makes statements on what is
>     >     actually done
>     >     >>> within a DetNet and the relationship with transport
>     protocols
>     >     that use detnet
>     >     >>> (which are in fact outside the scope of the DetNet WG). 
>     I'll
>     >     work with the
>     >     >>> authors and WG on this -- I see this change as
>     important, but
>     >     editorial in
>     >     >>> nature.
>     >     >>>
>     >     >>> - We have a perception issue with at least one member of the
>     >     TSV area on the
>     >     >>> meaning and more importantly, implication, of the term
>     "DetNet
>     >     >>> *Transport* sub-layer".  While I don't disagree that a good
>     >     portion of the IETF
>     >     >>> thinks transport protocol (UDP/TCP) when they hear
>     "transport"
>     >     >>> there are plenty others, particularly in the routing
>     area, who
>     >     understand that
>     >     >>> "transport" can refer to Transport Networks.  And Transport
>     >     Network is a well
>     >     >>> understood general industry term. The IETF even has a
>     bunch of
>     >     RFCs that
>     >     >>> relate to Transport networks.  This said, I think it
>     >     reasonable to go back to the
>     >     >>> DetNet WG and discuss changing the name of the "DetNet
>     >     Transport sub-
>     >     >>> layer"  to avoid the word "transport". -- BTW we made a
>     >     parallel change in
>     >     >>> the TEAS WG when producing RFC8453.
>     >     >>>
>     >     >>> See below for detail response in-line.
>     >     >>>
>     >     >>> On 11/19/2018 5:15 PM, Scharf, Michael wrote:
>     >     >>>> Lou,
>     >     >>>>
>     >     >>>>> --
>     >     >>>>> I wanted to take a step back from the multiple discussions
>     >     that were
>     >     >>>>> spawned by your review -- from a doc shepherd perspective,
>     >     and see
>     >     >>>>> where we are.   I know that the authors have sent a -09
>     >     version that
>     >     >>>>> addresses some, but not all issues.
>     >     >>>>>
>     >     >>>>>     From the exchanges I've seen, I think the key
>     remaining
>     >     issues are
>     >     >>>>> related to:
>     >     >>>>>
>     >     >>>>> (a) possibly introduction of congestion in the general
>     >     internet if
>     >     >>>>> packets were somehow to escape a detnet domain.  The
>     source
>     >     of this
>     >     >>>>> congestion would be inelastic traffic using DetNet or
>     due to
>     >     >>>>> congestion loss that is masked by PREOF.
>     >     >>>> These are two major issues that need to be addressed. Note
>     >     that it may not
>     >     >>> be sufficient just to add a section on operational and
>     deployment
>     >     >>> considerations. Also the existing text in the document will
>     >     need to get aligned
>     >     >>> to normative guidance on how to avoid a congestion collapse.
>     >     >>>> In -09, one example would be Section 3.1. "Primary goals
>     >     defining the
>     >     >>> DetNet QoS"
>     >     >>>>       Congestion protection operates by allocating
>     resources
>     >     along the path
>     >     >>>>       of a DetNet flow, e.g., buffer space or link
>     >     bandwidth.  Congestion
>     >     >>>>       protection greatly reduces, or even eliminates
>     >     entirely, packet loss
>     >     >>>>       due to output packet congestion within the
>     network, but
>     >     it can only
>     >     >>>>       be supplied to a DetNet flow that is limited at the
>     >     source to a
>     >     >>>>       maximum packet size and transmission rate.  Note that
>     >     congestion
>     >     >>>>       protection provided via congestion detection and
>     >     notification is
>     >     >>>>       explicitly excluded from consideration in DetNet,
>     as it
>     >     serves a
>     >     >>>>       different set of applications.
>     >     >>>> At least the last sentence would contradict a better
>     >     discussion of congestion
>     >     >>> in the document. For instance, it could just be removed. In
>     >     any case, the
>     >     >>> current wording in the last sentence is not correct, as the
>     >     IETF term for what is
>     >     >>> described in the last sentence is "congestion control".
>     >     >>>> Another example would be  Section 3.2.1.1. "Eliminate
>     >     congestion loss"
>     >     >>>>
>     >     >>>>       The primary means by which DetNet achieves its QoS
>     >     assurances is to
>     >     >>>>       reduce, or even completely eliminate, congestion
>     within
>     >     a DetNet node
>     >     >>>>       as a cause of packet loss. This can be achieved only
>     >     by the
>     >     >>>>       provision of sufficient buffer storage at each node
>     >     through the
>     >     >>>>       network to ensure that no packets are dropped due
>     to a
>     >     lack of buffer
>     >     >>>>       storage.  Note that a DetNet flow cannot be
>     throttled,
>     >     i.e., its
>     >     >>>>       transmission rate cannot be reduced via explicit
>     congestion
>     >     >>>>       notification.
>     >     >>>>
>     >     >>>> This section IMHO has to include a discussion of what
>     happens
>     >     in the (not
>     >     >>> expected) case that packets get dropped or that ECN
>     marks are
>     >     received. It is
>     >     >>> understood that this would not happen in normal
>     operation of a
>     >     DetNet
>     >     >>> network, but I believe just considering the error-free
>     >     operation of a DetNet
>     >     >>> network is not sufficient for this document. At least
>     for the
>     >     risk of traffic that
>     >     >>> may escape from a DetNet network is inherently not
>     sufficient
>     >     to assume that
>     >     >>> the DetNet network is always error-free.
>     >     >>>
>     >     >>> I think these are examples of text that needs to be
>     cleanup up
>     >     and to
>     >     >>> delineate what is done with a DetNet.
>     >     >>>
>     >     >>>
>     >     >>>> As a result, addressing my concerns will most likely
>     require
>     >     editing several
>     >     >>> parts of the document.
>     >     >>>> In addition, I'd like to emphasize that my review
>     comment "It
>     >     is surprising
>     >     >>> that there is hardly any discussion on network
>     robustness and
>     >     safety"
>     >     >>>
>     >     >>> I have no idea what you mean by safety here.  Can you
>     elaborate.
>     >     >>>
>     >     >>>
>     >     >>>> covers more than just inelastic traffic that escapes from a
>     >     DetNet network
>     >     >>> and masking of packet loss. Given that DetNet traffic may be
>     >     extremely critical
>     >     >>> traffic, I really wonder why the document doesn't emphasize
>     >     more the
>     >     >>> required robustness against failures *inside* the DetNet
>     >     network as well as
>     >     >>> counter-measures. But this is something the WG needs to
>     >     decide. As TSV-ART
>     >     >>> reviewer, I will be fine if the document clearly
>     describes how
>     >     the impact of
>     >     >>> failures will be isolated inside the DetNet network and will
>     >     not put the general
>     >     >>> Internet at risk.
>     >     >>>
>     >     >>> I agree - I think, the document should be clear on it's
>     scope and
>     >     >>> relationship to general internet usage.
>     >     >>>
>     >     >>>
>     >     >>>>> (b) The use of the term 'transport' in DetNet to refer to
>     >     what is
>     >     >>>>> basically a Traffic Engineered sub-network layer, such
>     as is
>     >     provided
>     >     >>>>> with MPLS-TE or Optical Transport Networks.
>     >     >>>> In the Internet architecture, the term 'transport'
>     refers to
>     >     Internet transport
>     >     >>> protocols. I doubt that the document can avoid
>     discussing the
>     >     implications of
>     >     >>> and interactions with Internet transport protocols such
>     as UDP
>     >     or TCP. As a
>     >     >>> result, I disagree that the document can use the term
>     >     'transport' to refer to
>     >     >>> traffic engineered sub-network layers.
>     >     >>>
>     >     >>> I think this is covered by my comment above.
>     >     >>>
>     >     >>>
>     >     >>>>    From a TSV-ART point of view, the document can
>     either only
>     >     use the term
>     >     >>> "transport" for Internet transport protocols and use another
>     >     term for sub-
>     >     >>> network layers (as handled in the *routing* area of the
>     IETF),
>     >     or the document
>     >     >>> has to clearly distinguish between the Internet transport
>     >     layer and other uses
>     >     >>> of the term "transport" and explain the overlap. I
>     believe the
>     >     former would be
>     >     >>> less confusing, but I will leave it up to the TSV ADs to
>     >     discuss terminology
>     >     >>> overlap in the IESG. As TSV-ART reviewer I insist that the
>     >     document uses the
>     >     >>> terms "transport layer" and "transport protocol" only when
>     >     referring to the
>     >     >>> Internet transport layer.
>     >     >>>
>     >     >>> I'm personally okay with a name change and even willing to
>     >     push this
>     >     >>> discussion within the WG, but as said above, "Transport
>     >     Network" is a
>     >     >>> generally understood industry term that is also used in RFCs
>     >     -- so we'll
>     >     >>> have to see what where WG consensus ends up.
>     >     >>>
>     >     >>>>> Do you have any other issues that that are critical to be
>     >     addressed
>     >     >>>>> before this work moves forward? If so which?
>     >     >>>> Regarding Section 4.4 I have already deferred the
>     discussion
>     >     to the IESG. The
>     >     >>> TSV-ART review comment is that the IESG needs to carefully
>     >     look at the
>     >     >>> concepts, terminology, and references in section 4.4.
>     >     >>>> Regarding my other comments, I acknowledge that -09 is
>     a step
>     >     forward. But
>     >     >>> given the cross-dependencies e.g. regarding terminology and
>     >     definitions, I will
>     >     >>> need to read the text completely once there is a
>     proposal how
>     >     to address my
>     >     >>> review. As noted in my review, I believe the document
>     must use
>     >     terminology
>     >     >>> clearly and consistently. As example, a statement in -09
>     such
>     >     as "Network
>     >     >>> nodes supporting DetNet flows have to implement some of
>     the DetNet
>     >     >>> capabilities (not necessarily all) in order to treat DetNet
>     >     flows such that their
>     >     >>> QoS requirements are met" is IMHO too vague. But in such
>     cases
>     >     it depends
>     >     >>> whether there is precise normative guidance elsewhere. And
>     >     this requires
>     >     >>> looking at the text as a whole.
>     >     >>>
>     >     >>> I think the next steps lie with me and the WG. We'll let you
>     >     know once
>     >     >>> there is a new version.  Of course, you can also
>     contribute to
>     >     the WG
>     >     >>> discussion on the topic.
>     >     >>>
>     >     >>> Thanks,
>     >     >>>
>     >     >>> Lou
>     >     >>>
>     >     >>>> Best regards
>     >     >>>>
>     >     >>>> Michael
>     >     >>>>
>     >     >>>>
>     >     >>>>
>     >     >>>>> Thank you,
>     >     >>>>>
>     >     >>>>> Lou
>     >     >>>>>
>     >     >>>>> On 9/28/2018 6:24 PM, Michael Scharf wrote:
>     >     >>>>>> Reviewer: Michael Scharf
>     >     >>>>>> Review result: Ready with Issues
>     >     >>>>>>
>     >     >>>>>> The document "Deterministic Networking Architecture"
>     >     >>>>>> (draft-ietf-detnet-architecture-08) defines an overall
>     >     framework for
>     >     >>>>>> Deterministic Networking.
>     >     >>>>>>
>     >     >>>>>> As TSV-ART reviewer, I believe that this document has
>     issues as
>     >     >>>>> detailed below.
>     >     >>>>>> Michael
>     >     >>>>>>
>     >     >>>>>> Major issues:
>     >     >>>>>>
>     >     >>>>>> * It seems that DetNet cannot easily be deployed in the
>     >     Internet
>     >     >>>>> without
>     >     >>>>>> additional means. Thus, for a baseline document, one
>     could
>     >     expect
>     >     >>>>> some
>     >     >>>>>> explanation on the requirements of deploying DetNet in a
>     >     network.
>     >     >>>>> DetNet
>     >     >>>>>> basically requires support in (almost) all network
>     devices
>     >     >>>>> transporting DetNet
>     >     >>>>>> traffic. That assumption should be explicitly spelt out
>     >     early in the
>     >     >>>>> document,
>     >     >>>>>> e.g., in the introduction. There also needs to be an
>     explicit
>     >     >>>>> discussion of the
>     >     >>>>>> implications if not the whole network is aware of or
>     >     supports DetNet.
>     >     >>>>> There is
>     >     >>>>>> some text in Section 4.2.2 and Section 4.3.3, but I
>     believe
>     >     >>>>> additional explicit
>     >     >>>>>> discussion is needed at a prominant place. For instance,
>     >     can use of
>     >     >>>>> DetNet do
>     >     >>>>>> harm to parts of a network not supporting DetNet? As
>     a side
>     >     note,
>     >     >>>>> when TCPM
>     >     >>>>>> published RFC 8257, the following disclaimer was added:
>     >     "DCTCP, as
>     >     >>>>> described in
>     >     >>>>>> this specification, is applicable to deployments in
>     controlled
>     >     >>>>> environments
>     >     >>>>>> like data centers, but it must not be deployed over
>     the public
>     >     >>>>> Internet without
>     >     >>>>>> additional measures." I wonder if a similar disclaimer is
>     >     needed for
>     >     >>>>> DetNet. If
>     >     >>>>>> there is an implicit assumption that DetNet will  be
>     used in
>     >     >>>>> homogenous
>     >     >>>>>> environments with mostly DetNet-aware devices within
>     the same
>     >     >>>>> organization,
>     >     >>>>>> such an assumption should be made explicit.
>     >     >>>>>>
>     >     >>>>>> * It is surprising that there is hardly any discussion on
>     >     network
>     >     >>>>> robustness
>     >     >>>>>> and safety; this probably also relates to security. For
>     >     instance,
>     >     >>>>>> misconfiguration or errors of functions performing packet
>     >     replication
>     >     >>>>> could
>     >     >>>>>> severely and permantly congest a network and cause harm.
>     >     How does the
>     >     >>>>> DetNet
>     >     >>>>>> architecture ensure that a network stays fully
>     operational
>     >     e.g. if
>     >     >>>>> the topology
>     >     >>>>>> changes or there are equipment failures? Probably
>     this can
>     >     be solved
>     >     >>>>> by
>     >     >>>>>> implementations (e.g., dynamic control plane), but
>     why are
>     >     >>>>> corresponding
>     >     >>>>>> requirements not spelt out? Section 3.3.2 speculates that
>     >     filters and
>     >     >>>>> policers
>     >     >>>>>> can help, and that may be true, but that probably
>     still assumes
>     >     >>>>> consistently
>     >     >>>>>> and correctly configured (and well-behaving) devices. And
>     >     Section
>     >     >>>>> 3.3.2 is
>     >     >>>>>> vague and mentions a "infinite variety of possible
>     >     failures" without
>     >     >>>>> stating
>     >     >>>>>> any requirements or recommendations. There may be further
>     >     solutions,
>     >     >>>>> such as
>     >     >>>>>> circuit breakers and the like. Why are such topics not
>     >     discussed?
>     >     >>>>>>
>     >     >>>>>> * Somewhat related, the document only looks at impact of
>     >     failures to
>     >     >>>>> the QoS of
>     >     >>>>>> DetNet traffic. What is missing is a discussion how to
>     >     protect non-
>     >     >>>>> DetNet parts
>     >     >>>>>> of a network from any harm caused by DetNet mechanisms.
>     >     Solutions to
>     >     >>>>> this
>     >     >>>>>> probably exist. But why is the impact on non-DetNet
>     traffic
>     >     (e.g., in
>     >     >>>>> case of
>     >     >>>>>> topology changes or failures of DetNet functions) not
>     >     discussed at
>     >     >>>>> all in the
>     >     >>>>>> document?
>     >     >>>>>>
>     >     >>>>>> * Regarding security, an architecture like DetNet
>     probably
>     >     requires
>     >     >>>>> that only
>     >     >>>>>> authenticated and authorized end systems have access
>     to the
>     >     data
>     >     >>>>> plane. The
>     >     >>>>>> security considerations only briefly mention the control
>     >     aspect ("the
>     >     >>>>>> authentication and authorization of the controlling
>     systems").
>     >     >>>>>>
>     >     >>>>>> * For an architecture document, the lack of clarity and
>     >     consistency
>     >     >>>>> regarding
>     >     >>>>>> terminology is concerning. This specifically applies
>     to the
>     >     case of
>     >     >>>>> incomplete
>     >     >>>>>> networks (as per Section 4.2.2 and 4.3.3) that
>     include "DetNet-
>     >     >>>>> unaware nodes".
>     >     >>>>>> The document introduces terms such as "DetNet
>     intermediate
>     >     nodes" but
>     >     >>>>> then
>     >     >>>>>> repeatedly uses generic terms such as "node" or "hop"
>     that may
>     >     >>>>> include
>     >     >>>>>> DetNet-unaware nodes. For instance, for incomplete
>     networks, a
>     >     >>>>> sentence such as
>     >     >>>>>> "The primary means by which DetNet achieves its QoS
>     >     assurances is to
>     >     >>>>> reduce, or
>     >     >>>>>> even completely eliminate, congestion within a node as a
>     >     cause of
>     >     >>>>> packet loss"
>     >     >>>>>> seems to only apply to "DetNet transit nodes" but not
>     >     "DetNet-unaware
>     >     >>>>> nodes".
>     >     >>>>>> Similar ambiguity exist for other use of the terms "hop"
>     >     and "node",
>     >     >>>>> which may
>     >     >>>>>> or may not include DetNet-unaware nodes. It is
>     unclear why the
>     >     >>>>> document does
>     >     >>>>>> not consistently use the terminology introduced in
>     Section
>     >     2.1 in all
>     >     >>>>> sections
>     >     >>>>>> and clearly distinguishes cases with and without DetNet
>     >     support.
>     >     >>>>>>
>     >     >>>>>> * Section 4.4 refers to RFC 7426, which is an
>     informational
>     >     RFC on
>     >     >>>>> IRTF stream,
>     >     >>>>>> and the document uses the concepts introduced there
>     (e.g.,
>     >     "planes").
>     >     >>>>> This is
>     >     >>>>>> very confusing. First, an IETF Proposed Standard should
>     >     probably
>     >     >>>>> refer to
>     >     >>>>>> documents having IETF consensus. An example would be RFC
>     >     7491, albeit
>     >     >>>>> there is
>     >     >>>>>> other related work as well, e.g., in the TEAS WG. Second,
>     >     Section 4.4
>     >     >>>>> is by and
>     >     >>>>>> large decoupled from the rest of the document and not
>     >     specific to
>     >     >>>>> DetNet.
>     >     >>>>>> Neither do other sections of the document refer to
>     the concepts
>     >     >>>>> introduced in
>     >     >>>>>> Section 4.4, nor does Section 4.4 use the DetNet
>     terminology or
>     >     >>>>> discuss
>     >     >>>>>> applicability to DetNet. Section 4.4 even mentions
>     >     explicitly at the
>     >     >>>>> end that
>     >     >>>>>> it discusses aspects that are orthogonal to the DetNet
>     >     architecture.
>     >     >>>>> It is not
>     >     >>>>>> at all clear why Section 4.4 is in this document. Section
>     >     4.4 could
>     >     >>>>> be removed
>     >     >>>>>> from the document without impacting the rest of the
>     document.
>     >     >>>>>>
>     >     >>>>>> Minor issues:
>     >     >>>>>>
>     >     >>>>>> * Terminology "DetNet transport layer"
>     >     >>>>>>
>     >     >>>>>>       The term "transport layer" has a well-defined
>     meaning
>     >     in the IETF,
>     >     >>>>> e.g.
>     >     >>>>>>       originating from RFC 1122. While "transport"
>     and e.g.
>     >     "transport
>     >     >>>>> network" is
>     >     >>>>>>       used in the IETF for different technologies in
>     >     different areas, I
>     >     >>>>> think the
>     >     >>>>>>       term "transport layer" is typically understood to
>     >     refer to
>     >     >>>>> transport
>     >     >>>>>>       protocols such as TCP and UDP. As such, I
>     personally
>     >     find the term
>     >     >>>>> "DetNet
>     >     >>>>>>       transport layer" misleading and confusing. The
>     >     confusion is easy
>     >     >>>>> to see e.g.
>     >     >>>>>>       in Figure 4, where UDP (which is a transport
>     protocol
>     >     as per RFC
>     >     >>>>> 1122) sits
>     >     >>>>>>       on top of "transport".
>     >     >>>>>>
>     >     >>>>>>       Based on the document it also may be
>     >     solution/implementation
>     >     >>>>> specific whether
>     >     >>>>>>       the "DetNet transport layer" is actually a separate
>     >     protocol layer
>     >     >>>>> compared
>     >     >>>>>>       to the "DetNet service layer". Thus it is not clear
>     >     to me why the
>     >     >>>>> word
>     >     >>>>>>       "layer" has to be used, specifically in combination
>     >     "transport
>     >     >>>>> layer".
>     >     >>>>>>       To me as, the word "transport layer" (and
>     "transport
>     >     protocol")
>     >     >>>>> should be
>     >     >>>>>>       used for protocols defined in TSV area, consistent
>     >     with RFC 1122.
>     >     >>>>> But this is
>     >     >>>>>>       probably a question to be sorted out by the IESG.
>     >     >>>>>>
>     >     >>>>>> * Page 9
>     >     >>>>>>
>     >     >>>>>>        A DetNet node may have other resources requiring
>     >     allocation
>     >     >>>>> and/or
>     >     >>>>>>        scheduling,
>     >     >>>>>>
>     >     >>>>>>       This is just one of several examples for
>     inconsistent
>     >     use of
>     >     >>>>> terminology.
>     >     >>>>>>       What is a "DetNet node"? That term is not
>     introduced
>     >     in Section
>     >     >>>>> 2.1
>     >     >>>>>> * Page 14
>     >     >>>>>>
>     >     >>>>>>        A DetNet network supports the dedication of a high
>     >     proportion
>     >     >>>>> (e.g.
>     >     >>>>>>        75%) of the network bandwidth to DetNet flows.
>     >     >>>>>>
>     >     >>>>>>       The 75% value is not reasoned. What prevents using
>     >     99% of the
>     >     >>>>> bandwidth for
>     >     >>>>>>       DetNet traffic?
>     >     >>>>>>
>     >     >>>>>> * Page 15: Figure 2
>     >     >>>>>>
>     >     >>>>>>       If the term "transport layer" cannot be
>     avoided, the
>     >     labels in
>     >     >>>>> this figure
>     >     >>>>>>       should at least be expanded to "DetNet
>     transport layer".
>     >     >>>>>>
>     >     >>>>>> * Page 18: Figure 4
>     >     >>>>>>
>     >     >>>>>>       As already mentioned earlier, Figure 4 is
>     confusing.
>     >     UDP is a
>     >     >>>>> transport
>     >     >>>>>>       protocol. If the term "transport" cannot be
>     avoided,
>     >     the labels in
>     >     >>>>> this
>     >     >>>>>>       figure should at least be expanded to "DetNet
>     transport".
>     >     >>>>>>
>     >     >>>>>> * Page 23
>     >     >>>>>>
>     >     >>>>>>        If the source transmits less data than this limit
>     >     >>>>>>        allows, the unused resource such as link bandwidth
>     >     can be made
>     >     >>>>>>        available by the system to non-DetNet packets.
>     >     >>>>>>
>     >     >>>>>>       Could there be additional requirements on the
>     use of
>     >     unused
>     >     >>>>> resources by
>     >     >>>>>>       non-DetNet packets, e.g., regarding preemption?
>     I am just
>     >     >>>>> wondering... If
>     >     >>>>>>       that was possible, a statement like "... can be
>     made
>     >     available by
>     >     >>>>> the system
>     >     >>>>>>       to non-DetNet packets as long as all guarantees are
>     >     fulfilled"
>     >     >>>>> would be on
>     >     >>>>>>       the safe side, no?
>     >     >>>>>>
>     >     >>>>>> * Page 27:
>     >     >>>>>>
>     >     >>>>>>        DetNet achieves congestion protection and bounded
>     >     delivery
>     >     >>>>> latency by
>     >     >>>>>>        reserving bandwidth and buffer resources at every
>     >     hop along the
>     >     >>>>> path
>     >     >>>>>>        of the DetNet flow.
>     >     >>>>>>
>     >     >>>>>>       Why does this sentence use the word "hop"? As
>     far as
>     >     I understand,
>     >     >>>>> in DetNet
>     >     >>>>>>       bandwidth and buffer resources are reserved in each
>     >     DetNet
>     >     >>>>> intermediate node.
>     >     >>>>>>       If there were hops over IP routers not being DetNet
>     >     intermediate
>     >     >>>>> nodes, no
>     >     >>>>>>       resources would be reserved there. As per Section
>     >     4.3.3, it is
>     >     >>>>> possible to
>     >     >>>>>>       deploy DetNet this way. And obviously there can be
>     >     resource
>     >     >>>>> bottlenecks below
>     >     >>>>>>       IP, on devices that are not routers... So does
>     "hop"
>     >     here refer to
>     >     >>>>> IP router
>     >     >>>>>>       hops or also to devices not processing IP (or
>     IP/MPLS)?
>     >     >>>>>>
>     >     >>>>>> * Page 27:
>     >     >>>>>>
>     >     >>>>>>        Standard queuing and transmission selection
>     >     algorithms allow a
>     >     >>>>>>        central controller to compute the latency
>     >     contribution of each
>     >     >>>>>>        transit node to the end-to-end latency, ...
>     >     >>>>>>
>     >     >>>>>>       The text does not explain why a _central_
>     controller
>     >     is needed for
>     >     >>>>> this
>     >     >>>>>>       computation. Why would a distributed control plane
>     >     not be able to
>     >     >>>>> realize
>     >     >>>>>>       this computation. Isn't this
>     implementation-specific?
>     >     >>>>>>
>     >     >>>>>> * Page 32
>     >     >>>>>>
>     >     >>>>>>       To somebody who is not deeply familiar with
>     DetNet, it is
>     >     >>>>> impossible to parse
>     >     >>>>>>       the description of the examples in Section
>     4.7.3. For
>     >     instance,
>     >     >>>>> "VID +
>     >     >>>>>>       multicast MAC address" is not introduced. I think
>     >     this example
>     >     >>>>> must be
>     >     >>>>>>       expaned with additional context and explanation
>     to be
>     >     useful to
>     >     >>>>> readers.
>     >     >>>>>> * Page 34
>     >     >>>>>>
>     >     >>>>>>        There are three classes of information that a
>     >     central controller
>     >     >>>>> or
>     >     >>>>>>        distributed control plane needs to know that can
>     >     only be obtained
>     >     >>>>>>        from the end systems and/or nodes in the network.
>     >     >>>>>>
>     >     >>>>>>       Wouldn't it be sufficient to state "Provisioning of
>     >     DetNet
>     >     >>>>> requires knowledge
>     >     >>>>>>       about ...". Does it matter in this context
>     whether the
>     >     >>>>> provisioning is done
>     >     >>>>>>       by a central controller or a distributed control
>     >     plane? For
>     >     >>>>> instance, could
>     >     >>>>>>       the same paragraph also apply to a network that
>     uses
>     >     _multiple_
>     >     >>>>> central
>     >     >>>>>>       controllers, or hybrid combinations of central
>     >     controllers and
>     >     >>>>> distributed
>     >     >>>>>>       control planes? In general, an architecture
>     document
>     >     should be
>     >     >>>>> agnostic to
>     >     >>>>>>       implementation aspects unless there is a specific
>     >     need. In this
>     >     >>>>> specific
>     >     >>>>>>       case, I fail to see a need to discuss the
>     realization
>     >     of the
>     >     >>>>> control plane of
>     >     >>>>>>       a network.
>     >     >>>>>>
>     >     >>>>>> Editorial nits:
>     >     >>>>>>
>     >     >>>>>> * Page 9:
>     >     >>>>>>
>     >     >>>>>>        The low-level mechanisms described in Section 4.5
>     >     provide the
>     >     >>>>>>        necessary regulation of transmissions by an end
>     >     system or
>     >     >>>>>>        intermediate node to provide congestion
>     protection.  The
>     >     >>>>> allocation
>     >     >>>>>>        of the bandwidth and buffers for a DetNet flow
>     requires
>     >     >>>>> provisioning
>     >     >>>>>>        A DetNet node may have other resources requiring
>     >     allocation
>     >     >>>>> and/or
>     >     >>>>>>        scheduling, that might otherwise be
>     over-subscribed
>     >     and trigger
>     >     >>>>> the
>     >     >>>>>>        rejection of a reservation.
>     >     >>>>>>
>     >     >>>>>>       Probably a full stop is missing after
>     "provisioning".
>     >     >>>>>>
>     >     >>>>>> * Page 11: "... along separate (disjoint non-SRLG)
>     paths ..."
>     >     >>>>>>
>     >     >>>>>>       I find this confusing. I would understand e.g.
>     "along
>     >     separate
>     >     >>>>>>       (SRLG-disjoint) paths".
>     >     >>>>>>
>     >     >>>>>> * Page 34:
>     >     >>>>>>
>     >     >>>>>>        When using a peer-
>     >     >>>>>>        to-peer control plane, some of this
>     information may
>     >     be required
>     >     >>>>> by a
>     >     >>>>>>        system's neighbors in the network.
>     >     >>>>>>
>     >     >>>>>>       Would "acquired" be a better term?
>     >     >>>>>>
>     >     >>>>>> * Page 34:
>     >     >>>>>>
>     >     >>>>>>        o  The identity of the system's neighbors, and the
>     >     >>>>> characteristics of
>     >     >>>>>>           the link(s) between the systems, including the
>     >     length (in
>     >     >>>>>>           nanoseconds) of the link(s).
>     >     >>>>>>
>     >     >>>>>>       "Latency" or "delay" would probably be a better
>     terms
>     >     if the value
>     >     >>>>> is
>     >     >>>>>>       measured in nanoseconds.
>     >     >>>>>>
>     >     >>>>>> * Page 35:
>     >     >>>>>>
>     >     >>>>>>        DetNet is provides a Quality of Service (QoS), and
>     >     as such, does
>     >     >>>>> not
>     >     >>>>>>        directly raise any new privacy considerations.
>     >     >>>>>>
>     >     >>>>>>       Broken sentence
>     >     >>>>>>
>     >     >>>>>> * Please expand acronyms on first use (e.g., OTN)
>     >     >>>>>>
>     >     >>>>>>
>     >     >>>> _______________________________________________
>     >     >>>> detnet mailing list
>     >     >>>> [e-mail] <mailto:[e-mail]>
>     <mailto:[e-mail] <mailto:[e-mail]>>
>     >     >>>> https://www.ietf.org/mailman/listinfo/detnet
>     >
>     >     _______________________________________________
>     >     detnet mailing list
>     > [e-mail] <mailto:[e-mail]> <mailto:[e-mail]
>     <mailto:[e-mail]>>
>     > https://www.ietf.org/mailman/listinfo/detnet
>     >
>