Re: [tsvwg] Benjamin Kaduk's No Objection on draft-ietf-tsvwg-le-phb-09: (with COMMENT)

[Roland Bless <[e-mail]>]

Hi Benjamin,

On 17.02.19 at 20:24 Benjamin Kaduk wrote:
> ----------------------------------------------------------------------
> COMMENT:
> ----------------------------------------------------------------------
> 
> This document is generally in fine shape, and has some good discussion
> in it.  I basically just have some editorial suggestions.
> 
> As one general note, the text sometimes seems to present a mixed story,
> for example in the case of "downgrading" traffic from other PHBs.  We're
> told to in general not do this instead of dropping traffic, but on the
> other hand an example use of the LE PHB is for downgrading traffic from
> some other PHB (but only when it does not violate the operational
> objectives of that other PHB).  Greater clarity on who is authorized to
> decide to downgrade, and in what cases it makes sense, could be useful.

Ok, I'll re-read the text and try to add a bit of clarifying text.

> In a similar vein, the text sometimes suggests a dichotomy between use
> of the LE PHB for "preemptable" traffic vs. as a "scavenger" service
> class, and at other times suggests that these usages are identical.  But
> those are, of course, editorial matters.

Thanks for pointing that out. I think you are right, but it
is the very nature of the LE: it should scavenge residual capacity,
but it must be preemptable by BE in case it needs more resources,
so it has both facets.


> Abstract
> 
>                                         The primary objective of this LE
>    PHB is to protect best-effort (BE) traffic (packets forwarded with
>    the default PHB) from LE traffic in congestion situations, i.e., when
>    resources become scarce, best-effort traffic has precedence over LE
>    traffic and may preempt it.  Alternatively, packets forwarded by the
>    LE PHB can be associated with a scavenger service class, i.e., they
>    scavenge otherwise unused resources only.  [...]
> 
> It seems like "preemptable" and "scavenger" are being deliberately
> conflated, but are not necessarily the same.

The scavenger part came in somewhat late and that may have caused
the confusion by the conflation. I'll try to clarify this a bit more.

> Section 3
> 
>    (e.g., if a transport connection fails due to timing out, the
>    application may try several times to re-establish the transport
>    connection in order to resume the application session before finally
> 
> There was some directorate review traffic suggesting that further
> qualifications about the retries; I do see such qualifying statemnets in
> the next paragraph, so maybe there is no big need to do so here as well..

Yep.

>    Use of the LE PHB might assist a network operator in moving certain
>    kinds of traffic or users to off-peak times.  Alternatively, or in
>    addition, packets can be designated for the LE PHB when the goal is
>    to protect all other packet traffic from competition with the LE
> 
> Is it "alternatively" or "in addition" -- can it really be both at the
> same time?  (I suppose the intent is that different operators could
> apply different policies?)

This formulation is actually taken unchanged from the original RFC 3662,
but I may rephrase it. I think it can be "in addition", because
it may belong to different traffic aggregates. The background of this
stems IMHO from the fact that in former times, some operators throttled
or blocked peer-to-peer file sharing traffic. A better option would be
to let users mark it as LE and simply accept it, because it does no
harm to other traffic.

> Section 9
> 
> Is there a section reference in RFC 3754 to point us to?

I don't think so, because the whole RFC discusses Multicast and Diffserv.

> (Also, the RFC Editor will probably put a comma after "Basically".)

Done. :-)

>    Several forwarding error correction coding schemes such as fountain
>    codes (e.g., [RFC5053]) allow reliable data delivery even in
> 
> I'm used to seeing "forward error correction"; is "forwarding error
> correction" also an acceptabale usage?

Good catch, will use Forward Error Correction.

>    While the resource contention problem caused by multicast packet
>    replication is also true for other Diffserv PHBs, LE forwarding is
>    special, because often it is assumed that LE packets get only
> 
> nit: s/get only/only get/

Done.

>    forwarded in case of available resources at the output ports.  The
>    previously mentioned redundancy data traffic could nicely use the
>    varying available residual bandwidth being utilized the by LE PHB,
>    but only if the previously specific requirements in the internal
>    implementation of the network devices are considered.
> 
> I'm not sure how to interpret "previously specific requirements", here.
> Was it intended to be "previously specified requirements"?

Ok, then I need to state that more explicitly: it refers to the
replication condition in the preceding paragraph.

> Section 12
> 
> As per the GenART review, updating the draft in missref is a bit weird,
> but we can probably leave it to the responsible AD and RFC Editor to
> decide whether to retain the "Updates" relationship or directly effect
> the needed changes.

Spencer already replied to this.

Best regards,
 Roland