Re: [eman] EMAN-REQ: the notion of importance

[Juergen Quittek <[e-mail]>]

Hi Mouli,

I agree in general, if you reason about any kind of network management.
However, we are designing a standard for energy management.

And in this scope "power reduction priority"
(or how we would call it descriptively)
is a rather clear mechanism that you could use
for "importance-based power reduction by assigning
lower priorities to less important entities.
However you could use it also for power reduction
policies that consider other constraints than "importance".

Thanks,
    Juergen


On 02.03.12 07:33, "Mouli Chandramouli (moulchan)" <[e-mail]>
wrote:

>Power priority or Power shedding are focused on a single use case;
>whereas a concept of importance  is more general.
> 
>It is another tag (post-it to borrow the term coined by Juergen S.);
>which can be useful other use cases.
> 
>Thanks
>Mouli
> 
> 
>From: Benoit Claise (bclaise)
>Sent: Friday, March 02, 2012 1:01 AM
>To: Juergen Quittek
>Cc: Brad Schoening; Rolf Winter; John Parello (jparello); Mouli
>Chandramouli (moulchan); Ira McDonald; eman mailing list
>Subject: Re: [eman] EMAN-REQ: the notion of importance
>
>
> 
>Hi Juergen,
>
>Taking back your words:
>I would like to standardize a mechanism, in this case the power
>downpriority.  That's what standards do.  I do not see reason to limitthe
>application of the mechanism (power down priority) to a singleUse case
>(power down less business relevant devices first).On one side, you want a
>mechanism not limited to a single case (which I agree with).
>On the other side, you're ready to call it "power shedding", which limit
>this to a single use case.
>
>To leads me to think that the generic term "importance" was maybe not
>perfect, but actually better as it took into account more use cases...
>
>Regards, Benoit.
>
>
>Hi Brad, Thanks for this hint.  Being not a native user I thought about
>poweringdown to a lower power state, not about powering off.  But this
>doesn'tseem to be the way the term is commonly used.  Power shedding
>appears tobe much better suited. Thanks,    Juergen  On 01.03.12 17:25,
>"Brad Schoening" <[e-mail]> <mailto:[e-mail]> wrote:
>Juergen, Power shedding is probably a more accurate term for the use
>cases here forpriority/importance than just simply power down.  There are
>many things ina commercial setting that can be turned down, but not
>necessarily off.Things such as variable speed fans, battery chargers,
>etc.   On 3/1/12 7:53 AM, "Juergen Quittek" <[e-mail]>
><mailto:[e-mail]> wrote: Hi Benoit, I would like to standardize
>a mechanism, in this case the power downpriority.  That's what standards
>do.  I do not see reason to limitthe application of the mechanism (power
>down priority) to a singleUse case (power down less business relevant
>devices first). Why should the IETF do so?  Our task is to define useful
>mechanisms.I do not like excluding other use cases.  Take for example a
>networkwith two kinds of devices: - a few devices consuming a lot of
>energy and having high energy   saving potential - a huge amount of
>devices with low power demand and very little   Power saving potential
>when turned to sleep mode. Even if the business importance of the few
>major power consumersis higher than the business importance of the many
>small devices,an energy manager may decide to achieve its power saving
>objectiveseasier by powering down a just few main energy consumers
>instead ofpowering down myriads of small devices that only
>marginallycontribute to energy saving. We can't foresee constraints to be
>considered for powering downDevices.  Giving the operator a "priority"
>allows the operatorto implement any scheme, may it be based on importance
>or mot. Thanks,   Juergen  On 01.03.12 16:03, "Benoit Claise"
><[e-mail]> <mailto:[e-mail]> wrote:       Juergen,
>Rolf, John    Looking at Rolf's feedback:      I thought this is what you
>refer to as importance. If you have toswitchsomething off because you
>cannot power all devices and you have todecidebetween 911 services or the
>phone in the janitors office, the prioritywill tell you. So this is EMAN
>and I think we can say that, whateverthisobject means it has to do with
>energy and I agree with your example thatit helps you to decide what to
>power-off first in case you need to/wantto. If this is what importance
>means (I personally would still call itsomething less ambiguous, but if
>we describe it better I am fine withit)I think it is something relevant.
>But you were referring to other usecases. Care to share more?     Would
>you guys be happier with a compromise such as "business   importance",
>"context importance" or "Energy Management Importance"?    Expanding on
>Juergen's proposal:   OLD:      5.1.3. Power-down priority   The standard
>must provide means for retrieving and reporting  power priorities of
>powered entities. Power-down priorities indicate  an order in which
>powered entities should be switched to lower power  states in case lower
>power states are desired.     NEW:      5.1.3. xxxxx   The standard must
>provide means for ranking devices in the context  of a site or
>deployment, indicating which devices are more critical  to the operation.
>The value is useful during peak demand whendeciding  which devices could
>be turned off. A ranking of devices gives an  operator or control system
>a way to determine which devices should  receive power or could be turned
>off for cost savings during peak  hours of operation. In other words, if
>an operator is asked to turnoff  devices during a certain period, xxxx
>indicates an order in whichpowered  entities should be switched to lower
>power states.  Regarding your role proposal 5.1.2, I believe it's fine.
>Regards, Benoit (as a contributor)       Dear all, The requirements draft
>is the first one to be agreed on.We can do this without having to deal
>with all detailsthat the framework and the MIB modules can solve. In the
>current version draft-ietf-eman-requirements-05 thereis a requirement OLD
> 5.1.2.  Context information on powered entities   The energy management
>standard must provide means for retrieving and  reporting context
>information on powered entities, for example, tags  associated with a
>powered entity that indicate the powered entity's  role, or importance.
>Seeing the ongoing discussion I suggest separating "role"
>and"importance"and moving from the fuzzy term "importance" to "power-down
>priority".This would look like the following: NEW  5.1.2.  Context
>information on powered entities   The standard must provide means for
>retrieving and reporting context  information on powered entities, for
>example, tags associated with a  powered entity that indicate the powered
>entity's role.   5.1.3. Power-down priority   The standard must provide
>means for retrieving and reporting  power priorities of powered entities.
>Power-down priorities indicate  an order in which powered entities should
>be switched to lower power  states in case lower power states are
>desired. I think that the proposed requirement 5.1.3 covers Rolf's
>requirements  for accurate naming and John's requirements for the
>functionality hecalls "importance". Thanks,   Juergen  On 29.02.12 10:02,
>"Rolf Winter" <[e-mail]>
><mailto:[e-mail]><mailto:[e-mail]>
><mailto:[e-mail]> wrote:          Hey John, I am not asking
>for an IANA registry but a good description andjustification of
>importance. For most requirements it is just naturallyclear to have them
>such as having the ability to monitor power states.Nojustification needed
>in my opinion. Then a half sentences in thedocumentrequires something
>that is called "importance". Here I see a need for adescription and
>justification because it means different things todifferent people. BTW,
>I don't think that priority means the order in which devices needtobe
>powered up. It certainly doesn’t mean that in the PoE context: "This
>object controls the priority of the port from the pointof view of a power
>management algorithm.  The priority thatis set by this variable could be
>used by a control mechanismthat prevents over current situations by
>disconnecting firstports with lower power priority.  Ports that connect
>devicescritical to the operation of the network - like the E911telephones
>ports - should be set to higher priority." I thought this is what you
>refer to as importance. If you have to switchsomething off because you
>cannot power all devices and you have todecidebetween 911 services or the
>phone in the janitors office, the prioritywill tell you. So this is EMAN
>and I think we can say that, whateverthisobject means it has to do with
>energy and I agree with your example thatit helps you to decide what to
>power-off first in case you need to/wantto. If this is what importance
>means (I personally would still call itsomething less ambiguous, but if
>we describe it better I am fine withit)I think it is something relevant.
>But you were referring to other usecases. Care to share more? Best, Rolf
>NEC Europe Limited | Registered Office: NEC House, 1 Victoria Road,London
>W3 6BL | Registered in England 2832014             -----Original
>Message-----From: John Parello (jparello)
>[mailto:[e-mail]]Sent: Dienstag, 28. Februar 2012 20:05To: Rolf
>Winter; Mouli Chandramouli (moulchan); Ira McDonald; BradSchoeningCc:
>eman mailing listSubject: RE: [eman] EMAN-REQ: the notion of importance
>Hi Rolf, I used the terms in the email - it's defined in the
>framework,definitions and MIB.  I'm not just throwing terms out I'm
>trying tohelp to show *you* the difference in the email text. So let's
>focus onthe problem not try to discredit my word selection and
>transitivelymy premise in the drafts. On to the concept you're not
>seeing. Here's an example of the different concepts. Priority is
>ordering(precedence) like boot ordering,   while importance is
>context(significance). Example: So say I have devices on my trading floor
>and it is completely poweredoff. I may have to power  them up in a
>certain order based on prioritybut once they are up their running
>importance is different. (PRIORITY)Network ServicesFile ServicesSoftware
>/ Application Repository servers Database Servers ClientsAccess Lobby
>Phones Trading Phones Once they are running the importance to the
>business is different andcould be (IMPORTANCE)Network Services
>(90-100)Trading Phones  (80-90)File Services (70-80)Databases Servers
>(60-80)Client Access (30-50)Lobby Phones (10-30)Software / Application
>Repository Servers (1-20) The former is precedence the latter is
>significance.  Since priority isalready used in the PoE world for this I
>used "importance" todistinguish the concepts. Especially since the word
>priority us usedfor an action or process more times than for a device or
>thing. Sopriority IMO seemed more natural to the process or power versus
>adescription of the device. Simply put importance is needed to know what
>you can power off duringpeak demand (but not solely that's just one very
>major use case) BTW Notice my use of a "fuzzy"  name space for the device
>roles andimportance. Not all data needs IANA registry to be useful. So
>"fuzzy"does not equal bad. Site defined guided data is extremely useful.
>I've used importance with nearly a dozen EnMS vendors and scores
>ofvendors  and it's been easy to explain versus PoE priority. Happy
>toshow a running system if that clears it up. Suggest any new word
>youlike for the glossary and happy to discuss and select one but
>let'smake sure the concepts are retained. A bit shocked this is being
>debated for re-justification though as  Ifirst presented at IETF-78 and
>it's been in the drafts since then. To the Chairs: We need more input in
>this WG from EnMS vendors and BMSvendors because personally, dealing with
>over 100 vendors in acommunity of developers who use these concepts
>daily, I'm finding thoseactively participating in the group woefully not
>representative ofproblem space at all. We need more diverse input because
>these conceptsare in common use and a call for re-justification at this
>pointhighlights that weakness. Perhaps a demo of existing EnMS' to help
>educate the WG? Jp  -----Original Message-----From: [e-mail]
>[mailto:[e-mail]] On Behalf OfRolf WinterSent: Tuesday,
>February 28, 2012 1:16 AMTo: Mouli Chandramouli (moulchan); Ira McDonald;
>Brad SchoeningCc: eman mailing listSubject: Re: [eman] EMAN-REQ: the
>notion of importance Well let me make myself clearer then. You said:
>"Given the precedence of use of priority in other IETF MIBs,I think the
>value of importance is clearly illustrated." I disagreehere because some
>proponents of importance state that "Prioritydescribes precedence while
>importance describes significance. Those aretwo different concepts.". If
>that indeed is the case then youconclusion seems wrong. If priority !=
>importance then we shouldclearly describe what importance is. I think
>saying importance ==significance doesn't do the job. It is just a
>substitute of the wordusing a thesaurus but not a definition of how this
>is used and why thisis a requirement. But please go ahead and come
>forward with a gooddefinition of it and a good justification of it as a
>requirement. Wecan more concretely discuss about it then. Best, Rolf
>NEC Europe Limited | Registered Office: NEC House, 1 Victoria Road,London
>W3 6BL | Registered in England 2832014               -----Original
>Message-----From: Mouli Chandramouli (moulchan)
>[mailto:[e-mail]]Sent: Dienstag, 28. Februar 2012 10:02To: Rolf
>Winter; Ira McDonald; Brad SchoeningCc: eman mailing listSubject: RE:
>[eman] EMAN-REQ: the notion of importance Rolf, I do not know what you
>disagree on. Initially, some folks jumped on the bandwagon it is not
>useful inEnergy Management.And then a clear example of a similar term
>from the IETF PoE MIB wasshown. Now the question is definition of the
>term. I had mentioned in my email, that if it is a question of a
>clearerdefinition of the term, that can be provided. ThanksMouli
>-----Original Message-----From: Rolf Winter
>[mailto:[e-mail]]Sent: Tuesday, February 28, 2012 2:05 PMTo:
>Mouli Chandramouli (moulchan); Ira McDonald; Brad SchoeningCc: eman
>mailing listSubject: RE: [eman] EMAN-REQ: the notion of importance Mouli,
>I disagree. There are people on the list that seem to disagree
>thatimportance and priority are the same concept. Just the word
> importance             is utterly confusing. It could relate to
>security, cost,power-up orpower-down priority etc. Somebody mentioned PoE
>and there I agree itis clearly defined. Importance is not. Let us first
>clearly define            how             it is used, then let’s make a
>requirement out of it in casethe WGfeels it should be. And let us not
>forget to make clear what it meansin the context of EMAN. Best, Rolf  NEC
>Europe Limited | Registered Office: NEC House, 1 Victoria Road,London W3
>6BL | Registered in England 2832014                 -----Original
>Message-----From: [e-mail] [mailto:[e-mail]] On
>             Behalf                Of Mouli Chandramouli (moulchan)Sent:
>Dienstag, 28. Februar 2012 06:57To: Ira McDonald; Brad SchoeningCc: eman
>mailing listSubject: Re: [eman] EMAN-REQ: the notion of importance Given
>the precedence of use of priority in other IETF MIBs, I thinkthe value of
>importance is clearly illustrated.   Regarding Role, it is not intended
>to be an IANA registry.  Thisconcept is already used by deployments.
>Should not be dismissed asnot useful.   If the question is – clearer
>description of these terms, in therequirements draft, it is possible to
>provide some text and also              how                these concepts
>can be useful.   Thanks Mouli   From: [e-mail]
>[mailto:[e-mail]] On              Behalf                Of
>Ira McDonaldSent: Monday, February 27, 2012 11:15 PMTo: Brad Schoening;
>Ira McDonaldCc: eman mailing listSubject: Re: [eman] EMAN-REQ: the notion
>of importance   Hi, Brad - good precedent - because it makes the
>"importance"machine readable (and therefore useful). But since EMAN (and
>many other IETF WGs) have consistently backed              away
>    from any standard definition of "role" (w/ behaviorsemantics that
>         are               predictable), a text string of "role" is
>useless (exceptinavendor- or site-specific manner - out-of-scope IMHO).
>And I suggest that the "universe of things" is too diverse to lenditself
>to an IANA registry of standard "role" keywords. Cheers,- Ira  Ira
>McDonald (Musician / Software Architect) Chair - LinuxFoundation Open
>Printing WG Secretary - IEEE-ISTO Printer WorkingGroup Co-Chair-
>IEEE-ISTO PWG IPP WG Co-Chair - TCG Trusted Mobility Solutions WGChair-
>TCG Embedded Systems Hardcopy SG IETF Designated Expert - IPP &Printer
>MIB Blue Roof Music/High North
>Inchttp://sites.google.com/site/blueroofmusic<http://sites.google.com/site
>/ <http://sites.google.com/site/blueroofmusic>b
><http://sites.google.com/site/blueroofmusic>l
><http://sites.google.com/site/blueroofmusic>ueroofmusic>
><http://sites.google.com/site/blueroofmusic><http://sites.google.com/site/
>blueroofmusic> 
><http://sites.google.com/site/blueroofmusic>http://sites.google.com/site/h
>ighnorthinc<http://sites.google.com/site/highnorthinc>
><http://sites.google.com/site/highnorthinc><http://sites.google.com/site/h
>ighnorthinc> 
><http://sites.google.com/site/highnorthinc>mailto:[e-mail]
>inter  579 Park Place  Saline, MI  48176  734-944-0094 Summer  PO
>     Box               221  Grand Marais, MI 49839  906-494-2434     On
>Mon, Feb 27, 2012 at 12:10 PM, Brad Schoening <[e-mail]>
><mailto:[e-mail]><mailto:[e-mail]>
><mailto:[e-mail]>wrote: Benoit,   There is a precedence for doing
>this on the device in the PoE MIB,rfc3621 which defines
>pethPsePortPowerPriority:   pethPsePortPowerPriority OBJECT-TYPE   SYNTAX
>INTEGER   {              critical(1),              high(2),
>low(3)    }   MAX-ACCESS read-write   STATUS current   DESCRIPTION
>"This object controls the priority of the port from the
>point                         of view of a power management algorithm.
>Thepriority              that                         is set by this
>variable could be used by acontrol              mechanism
>        that prevents over current situations bydisconnecting
> first                         ports with lower power priority.  Ports
>thatconnect              devices                         critical to the
>operation of the network - likethe E911        telephones ports - should
>be set to higher priority."   ::= { pethPsePortEntry 7 }     Brad
>Schoeninge: [e-mail] ⟐ m: 917-304-7190
>Redefining Storage Economics     From: Benoit Claise <[e-mail]>
><mailto:[e-mail]> <mailto:[e-mail]>
><mailto:[e-mail]>Date: Mon, 27 Feb 2012 05:17:24 -0600To: eman
>mailing list <[e-mail]> <mailto:[e-mail]>
><mailto:[e-mail]> <mailto:[e-mail]>Subject: [eman] EMAN-REQ:
>the notion of importance   Dear all, There is a discussion amongst the
>"EMAN requirements" authors rightnow about the notion of importance.We're
>trying to evaluate the requirements related to the
>"importance".                The current draft
>version<http://tools.ietf.org/html/draft-ietf-
><http://tools.ietf.org/html/draft-ietf-eman-requirements-05>
><http://tools.ietf.org/html/draft-ietf-eman-requirements-05>
><http://tools.ietf.org/html/draft-ietf-eman-requirements-05>
><http://tools.ietf.org/html/draft-ietf-eman-requirements-05>
>eman- <http://tools.ietf.org/html/draft-ietf-eman-requirements-05>
><http://tools.ietf.org/html/draft-ietf-eman-requirements-05>
><http://tools.ietf.org/html/draft-ietf-eman-requirements-05>
>requirements-05> 
><http://tools.ietf.org/html/draft-ietf-eman-requirements-05>  only
>mentions:  5.1.2.  Context information on powered entities   The energy
>management standard must provide means for retrieving              and
>              reporting context information on powered entities,
>forexample,              tags                  associated with a powered
>entity that indicate thepowered              entity's
>role, or importance.  So there are no justifications why the importance
>is required.The people who want this, please provide some more
>  text/justifications               Some extra questions:- Is this
>importance specific to EMAN or is this generic also fornon Energy
>Objects?- Importance is important related to ...? Regards, Benoit (as a
>contributor for the EMAN-REQ)
>_______________________________________________eman mailing
>[e-mail]://www.ietf.org/mailman/listinfo/eman
>_______________________________________________eman mailing
>[e-mail]://www.ietf.org/mailman/listinfo/eman
>_______________________________________________eman mailing
>[e-mail]://www.ietf.org/mailman/listinfo/eman
>_______________________________________________eman mailing
>[e-mail]://www.ietf.org/mailman/listinfo/eman
>
> 
>
> 
>
>