Re: [Teas-ns-dt] draft-nsdt-teas-transport-slice-definition-02-comments

Hi Eric,
I agree with you that from an operational perspective degradation of
service performance below a certain threshold is equal to the failure. But,
AFAIK, that is not the interpretation of "Failure" in "Mean Time Between
Failures" (MTBF). ITU-T, as I understand, interprets "failure of a service"
as the inability to communicate whereas "service degradation" is pointing
to the worsening of the quality of the service while still maintaining
communication between endpoints. If we adopt these interpretations, then
MTBF should be complemented or replaced with something that also reflects
the acceptable level of the service quality.
What do you think?

Regards,
Greg

On Mon, May 18, 2020 at 5:35 AM Eric Gray <eric.gray@ericsson.com> wrote:

> Greg,
>
>
>
>               In this context, “failure” includes any form of service loss.
>
>
>
>               Performance degradation beyond acceptability is definitely a
> form of service loss.
>
>
>
>               In some cases, this is explicitly forced by terminating the
> service – but (even when this is not the case) the fact that the service is
> no longer delivering on commitment means that the service provider is no
> longer entitled to claim that it is “available.”
>
>
>
>               If I were a service provider, given the choice between
> wasting resources in an attempt to mis-deliver a service (and possibly
> being “caught out” in an apparent effort to charge for it anyway) or taking
> the service down (making it obviously the case that the service is not
> available), I would elect to do the latter.
>
>
>
>               In traditional (or “legacy”) services, reaching a
> degradation point was reported as a service failure and service was not
> considered “restored” until the service performance had improved to the
> point where the threshold was no longer a factor – for some well-defined
> period of time.
>
>
>
> --
>
> Eric
>
>
>
> *From:* Teas-ns-dt <teas-ns-dt-bounces@ietf.org> *On Behalf Of *Greg
> Mirsky
> *Sent:* Friday, May 15, 2020 8:40 PM
> *To:* Eric Gray <eric.gray=40ericsson.com@dmarc.ietf.org>
> *Cc:* LUIS MIGUEL CONTRERAS MURILLO <
> luismiguel.contrerasmurillo@telefonica.com>; teas-ns-dt@ietf.org; Luis M.
> Contreras <contreras.ietf@gmail.com>; Jari Arkko <jari.arkko@ericsson.com>;
> Shunsuke Homma <shunsuke.homma.fp@hco.ntt.co.jp>; Kiran Makhijani <
> kiranm@futurewei.com>; Shunsuke Homma <s.homma0718@gmail.com>; Rokui,
> Reza (Nokia - CA/Ottawa) <reza.rokui@nokia.com>
> *Subject:* Re: [Teas-ns-dt]
> draft-nsdt-teas-transport-slice-definition-02-comments
>
>
>
> Hi Erik, et al.,
>
> thank you for all the comments and suggestions to strengthen the document.
> I want to touch on the use of the term "availability" (with updates you've
> proposed):
>
> o Availability: concerns with how often a service is lost (or degraded to
> the point of unacceptable performance) due to a fault, and how quickly the
> lost service network becomes available after a fault. The requirements are
> specified through mean time between failures (MTBF), and mean time to
> repair (MTTR) to acquire availability metrics.
>
> I agree with you that availability is affected by both network failures
> and network performance degradation below a certain threshold. Then, if I
> follow that definition, the way of measuring or calculating the
> availability noted in the second sentence does not appear to be accurate as
> it only includes failure and does not consider periods of performance
> degradation. Since this draft, as I understand, is intended to provide the
> definitions to be used throughout other TS drafts, perhaps leaving in the
> definition will be sufficient, while the availability
> measurement/calculation method will be outside the scope of the document.
>
> What do you think?
>
>
>
> Regards,
>
> Greg
>
>
>
> On Wed, May 13, 2020 at 6:41 AM Eric Gray <eric.gray=
> 40ericsson.com@dmarc.ietf.org> wrote:
>
> Authors/Editors of draft-nsdt-teas-transport-slice-definition-02:
>
>
>
>               Please consider the attached revised and annotated version
> of your -02 version of this draft.
>
>
>
>               This version shows markup for removing or modifying some
> highly contentious or questionable text and sections – including
> annotations as to why this text should be removed.
>
>
>
>               It also includes comments and questions about other text
> that may not be necessary, or should be reworded.
>
>
>
>               One section (section 4.1.1) is shown as removed, but Jari
> had made a couple of other suggestions about that text immediately prior to
> the last meeting.  I do not recall if there was any decision about these
> alternatives and I could not track down the meeting notes from that meeting.
>
>
>
>               A key observation about this draft is that it is supposed to
> be a definitions draft – i.e. – it defines the terminology we are using or
> expect to use in other drafts related to this work..
>
>
>
>               There is some danger in having the ambition to see
> “definition” as something we can stretch to encompass “specification” in
> this draft.
>
>
>
>               It is important also to realize that the parameters and
> objectives of transport slice services listed in this draft are explicitly
> intended to be representative examples, rather than an exhaustive list –
> hence it is not necessary to include any for which there is no consensus to
> include.
>
>
>
> --
>
> Eric
>
> --
> Teas-ns-dt mailing list
> Teas-ns-dt@ietf.org
> https://www.ietf.org/mailman/listinfo/teas-ns-dt
>
>