[FoLUG] Fwd: Re: [energytransition] Re: data centre energy efficiency

Lun 11 Dic 2017 08:06:39 CET

Mari interessa qualcuno.

Ciao.
Ansfans
---------- Messaggio inoltrato ----------
Da: "Gregor Erbach" <gregor.erbach@gmail.com>
Data: 10 Dic 2017 10:08
Oggetto: Re: [energytransition] Re: data centre energy efficiency
A: <energytransition@googlegroups.com>
Cc:

The Joint Research Centre has developed a code of conduct for energy
> efficiency in data centres, for more information
> https://e3p.jrc.ec.europa.eu/communities/data-centres-code-conduct
>
> 2017-12-09 22:56 GMT+01:00 Alex Sorokin <sorokin@interenergy.it>:
>
>> Just some ideas on potentials for energy savings in data-centres:
>>
>> UPS: Present UPS technology is designed to provide highest possible
>> power-supply reliability al lowest cost. Energy efficiency is not a design
>> criterion. 50% of the energy self-consumed by UPS’s can be saved by
>> adopting energy-efficient battery-chargers, smart-charging algorithms and
>> more efficient inverters, but the market demand for such energy efficient
>> UPS's is small.
>>
>> Cooling: Present day cooling of IT is highly inefficient / dispersive
>> since provided mostly by non-focused devices (such as cooling fans) and
>> since air-conditioning systems treat ALL the air inside a data-centre,
>> without distinguishing between hot and cold air-streams. Focused
>> heat-extraction only from those components which really need cooling (such
>> as microprocessors) allows to reduce dramatically (probably 80% of) related
>> energy needs. Passive cooling and so-called "free cooling" is seldom
>> adopted.
>>
>> Data-centre-operators design their systems and purchase equipment by
>> focusing all their efforts on service reliability. They care little about
>> energy efficiency.
>>
>> Alex
>>
>>
>> Il 09/12/2017 18:51, Francois-Xavier Chevallerau ha scritto:
>>
>>
>> Unless I’m missing something, I don't think that the fact that the
>> majority of electricity used by data centers is for cooling and
>> air-conditioning means that the IT doesn’t consume much after all, since
>> you cannot run the IT without cooling and air-conditioning. And I’m also
>> not sure about how it shows that there are high technical potentials for
>> efficiency improvements, unless you’re referring to efficiency improvement
>> concerning cooling and air-conditioning systems.
>>
>>
>> Le samedi 9 décembre 2017 15:29:57 UTC+1, Alex Sorokin a écrit :
>>>
>>> From energy audits done on data centres we found that only 17% of
>>> consumed electricity is used to power the IT.
>>>
>>> 25% is consumed by UPS back-up systems, 2% for lighting and the rest
>>> (56%) for cooling and air-conditioning.
>>>
>>> Technical potentials for efficiency improvements are therefore high.
>>> Alex
>>>
>>> Il 09/12/2017 00:04, Thierry Caminel ha scritto:
>>>
>>> Jonathan,
>>>
>>> The coordinator of the study I've mentioned is Anders S.G. Andrae, who
>>> is working for 15 years on LCA and energy efficiency of electronic devices:
>>> https://scholar.google.fr/citations?hl=en&user=sIXZ
>>> Y6MAAAAJ&view_op=list_works&sortby=title
>>>
>>> For the datacenters, my correspondents know your studies, but they have
>>> also access to some consumption data.  Their results are less optimistic
>>> than yours regarding power consumption and efficiency improvement. They
>>> told  me for example that datacenter consumption in UK increased by 9 to
>>> 10% per year between 2014 and 2016.
>>>
>>> I'm not specialist so I cannot evaluate these different studies, whose
>>> scope seems also different.  However, I think you are wrong to qualify the
>>> article as "not credible". It's normal that you think first that your own
>>> studies are 'more credible', but you should stay as a scientist and accept
>>> without a priori discordant results.
>>> Final study from The Shift Project should be published in a few weeks, I
>>> will publish it here and give you the contacts to the authors, so hopefully
>>> that will lead to useful exchanges and a more consistent view on that
>>> important topic.
>>>
>>> Regarding energy intensity, the paper mentioned my Tere ([Kovacic 2017])
>>> provide I think an interesting answer to the point you raise.
>>>
>>> Best,
>>> Thierry
>>>
>>>
>>> 2017-12-06 21:02 GMT+01:00 Jonathan Koomey <jgko...@gmail.com>:
>>>
>>>> This is one of my prime research areas, so I know for a fact that the
>>>> article you cite is not credible.
>>>>
>>>> This is the latest credible research on the topic for the US. Little
>>>> growth since 2007 and little projected to 2020.
>>>>
>>>> Shehabi, Arman, Sarah Josephine Smith, Dale A. Sartor, Richard E.
>>>> Brown, Magnus Herrlin, Jonathan G. Koomey, Eric R. Masanet, Nathaniel
>>>> Horner, Inês Lima Azevedo, and William Lintner. 2016. *United States
>>>> Data Center Energy Usage Report*. Berkeley, CA: Lawrence Berkeley
>>>> National Laboratory. LBNL-1005775.  June. [
>>>> https://eta.lbl.gov/publications/united-states-data-center-energy]
>>>>
>>>> Also, you should never point to Primary energy per unit of GDP as a
>>>> metric for energy intensity, because it conflates energy system losses and
>>>> end-use efficiency gains. Just because history doesn’t show decoupling (and
>>>> I’m not granting that it doesn’t because the PFU data from IIASA show that
>>>> it does) it’s a separate issue from whether decoupling is possible. Two
>>>> different questions.
>>>>
>>>> Here’s the link for the PFU data:
>>>>
>>>> http://www.iiasa.ac.at/web/home/research/researchPrograms/Tr
>>>> ansitionstoNewTechnologies/PFUDB.en.html
>>>>
>>>> According to these data, the final energy/unit of Gross world product,
>>>> measured using purchasing power parity, declined 0.8%/year from 1900 to
>>>> 2014 globally.  From 1995 to 2014, FE/GWP declined at a 1.83%/year rate.
>>>>
>>>> The energy supply loss factor (PE/FE) increased 0.2%/year from 1900 to
>>>> 2014, and 0.24%/year from 1995 to 2014.
>>>>
>>>> Jon
>>>>
>>>>
>>>>
>>>> On Dec 4, 2017, at 5:28 PM, Thierry Caminel <tcam...@gmail.com> wrote:
>>>>
>>>> Jonathan,
>>>>
>>>> The study I'm talking about is made by people having first-hand data on
>>>> data centers consumption and energy needed to build smartphones and
>>>> servers. And 15% now is coherent with 5 to 6% you mention in 2007.
>>>>
>>>> Here the result of the compilation of data so far:
>>>> <image.png>
>>>>
>>>> Column 'expected' comes from [Andres & al] paper.  Column 'actual'
>>>> comes from data collected by the authors.
>>>> I'll give you updated raw numbers and sources when I have them.
>>>>
>>>>
>>>> Anyway, the fact is that there's no sign of improvement of the energy
>>>> intensity of GDP.  It's the same rate since 1950. No sign of decoupling so
>>>> far at world level.  So there maybe some "conceptual reason" behind,
>>>> don't you think?
>>>>
>>>> <image.png>
>>>>
>>>> source: https://jancovici.com/en/energy-transition/energy-and-us/wha
>>>> t-is-energy-actually/
>>>>
>>>>
>>>> In fact there are I think several reasons, like:
>>>> - Jevon effect  (computers consume less, but there are many more)
>>>> - New usages (a bitcoin transaction requires thousands more energy
>>>> than a Visa one
>>>> <https://digiconomist.net/bitcoin-energy-consumption%20/> ; Netflix
>>>> consumes more than radio broadcasting ; .... )
>>>> - Decreasing EROI   (we need more energy to produce energy; it is
>>>> embedded in the products and services whose value is in GDP)
>>>> - Increasing energy consumption due to transportation (e-commerce is
>>>> not energy efficient)
>>>> - Increasing energy required to extract / refine material
>>>> - Diminishing returns of social system complexity (Robert Tainter
>>>> argument)
>>>> - Robert Gordon's 'contrary winds' :  'http://fortune.com/2016/02/02
>>>> /end-of-growth/
>>>> ... and probably others.
>>>>
>>>>
>>>> Best, Thierry
>>>>
>>>>
>>>>
>>>>
>>>> 2017-12-05 1:26 GMT+01:00 Jonathan Koomey <jgko...@gmail.com>:
>>>>
>>>>> This is absolutely not true.  There’s no conceptual reason why we
>>>>> can’t decouple energy use (and more particularly emissions) from economic
>>>>> activity.  Why many people on this list seem to think this is true is a bit
>>>>> mystifying to me.  Some activities are more energy and emissions intensive
>>>>> and some are less. If we shift to activities that are less emissions
>>>>> intensive, then decoupling happens. Also, the 2nd law efficiency the
>>>>> economy is around 5%, which means we can increase efficiency of using
>>>>> energy manyfold before we run into thermodynamic limits.
>>>>>
>>>>> Also, this analysis on IT electricity use is deeply flawed and should
>>>>> not be cited. No credible analyst would ever project IT electricity use out
>>>>> 15 years, and the people who wrote the analysis are industry folks without
>>>>> a deep background or history in electricity estimates in IT.
>>>>> <image.png>
>>>>> Total electricity use for IT globally is around 5-6% based on data
>>>>> circa 2007 (which is the latest decent estimate that has been done).
>>>>>
>>>>> Malmodin, Jens, Åsa Moberg, Dag Lundén, Göran Finnveden, and Nina
>>>>> Lövehagen. 2010. "Greenhouse Gas Emissions and Operational Electricity Use
>>>>> in the ICT and Entertainment & Media Sectors."  *Journal of
>>>>> Industrial Ecology*.  vol. 14, no. 5. pp. 770-790. [
>>>>> http://dx.doi.org/10.1111/j.1530-9290.2010.00278.x]
>>>>>
>>>>> Data centers used less than 2% of US electricity in 2014, and that
>>>>> total has been roughly flat. World data center electricity use probably
>>>>> grew a bit more but not by much.
>>>>>
>>>>> Shehabi, Arman, Sarah Josephine Smith, Dale A. Sartor, Richard E.
>>>>> Brown, Magnus Herrlin, Jonathan G. Koomey, Eric R. Masanet, Nathaniel
>>>>> Horner, Inês Lima Azevedo, and William Lintner. 2016. *United States
>>>>> Data Center Energy Usage Report*. Berkeley, CA: Lawrence Berkeley
>>>>> National Laboratory. LBNL-1005775.  June. [
>>>>> https://eta.lbl.gov/publications/united-states-data-center-energy]
>>>>>
>>>>> Bottom line: people have this weird idea that we can’t decouple, but
>>>>> there’s no conceptual reason why we can’t.
>>>>>
>>>>> People have this weird idea that IT uses a ton of electricity and that
>>>>> it’s growing rapidly. It doesn’t, it isn’t, and for the few % of
>>>>> electricity that we devote to it, IT delivers immense value.
>>>>>
>>>>> Jon
>>>>>
>>>>>
>>> --
>>> You received this message because you are subscribed to the Google
>>> Groups "Energy Transition Network" group.
>>> To unsubscribe from this group and stop receiving emails from it, send
>>> an email to energytransiti...@googlegroups.com.
>>> To post to this group, send email to energytr...@googlegroups.com.
>>> To view this discussion on the web visit https://groups.google.com/d/ms
>>> gid/energytransition/CAEpnQ4xt-QHWkfdcaGMfALhnJCfX1XL-h6iJg3
>>> MdBUQQEUwJ2g%40mail.gmail.com
>>> <https://groups.google.com/d/msgid/energytransition/CAEpnQ4xt-QHWkfdcaGMfALhnJCfX1XL-h6iJg3MdBUQQEUwJ2g%40mail.gmail.com?utm_medium=email&utm_source=footer>
>>> .
>>> For more options, visit https://groups.google.com/d/optout.
>>>
>>>
>>> --
>> You received this message because you are subscribed to the Google Groups
>> "Energy Transition Network" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to energytransition+unsubscribe@googlegroups.com.
>> To post to this group, send email to energytransition@googlegroups.com.
>> To view this discussion on the web visit https://groups.google.com/d/ms
>> gid/energytransition/5ced70f9-f5f2-4059-805e-b001533a1f8f%
>> 40googlegroups.com
>> <https://groups.google.com/d/msgid/energytransition/5ced70f9-f5f2-4059-805e-b001533a1f8f%40googlegroups.com?utm_medium=email&utm_source=footer>
>> .
>> For more options, visit https://groups.google.com/d/optout.
>>
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "Energy Transition Network" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to energytransition+unsubscribe@googlegroups.com.
>> To post to this group, send email to energytransition@googlegroups.com.
>> To view this discussion on the web visit https://groups.google.com/d/ms
>> gid/energytransition/0ab5189a-c3ac-d8b3-6913-2af5b1dd44d0%
>> 40interenergy.it
>> <https://groups.google.com/d/msgid/energytransition/0ab5189a-c3ac-d8b3-6913-2af5b1dd44d0%40interenergy.it?utm_medium=email&utm_source=footer>
>> .
>>
>> For more options, visit https://groups.google.com/d/optout.
>>
>
> --
> You received this message because you are subscribed to the Google Groups
> "Energy Transition Network" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to energytransition+unsubscribe@googlegroups.com.
> To post to this group, send email to energytransition@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/
> msgid/energytransition/CADAPWbF6MWDVghgM_%3D_2%2Bh0JV%2Ba_
> Toa8fGA9pEBSiJVed3cB6g%40mail.gmail.com
> <https://groups.google.com/d/msgid/energytransition/CADAPWbF6MWDVghgM_%3D_2%2Bh0JV%2Ba_Toa8fGA9pEBSiJVed3cB6g%40mail.gmail.com?utm_medium=email&utm_source=footer>
> .
> For more options, visit https://groups.google.com/d/optout.
>