Resultados 1 a 2 de 2
  1. #1
    WHT-BR Top Member
    Data de Ingresso
    Dec 2010
    Posts
    15,049

    [EN] "Data Center Productivity Will Be Measured In Kittens Per KWh"

    There’s a new measure of data center efficiency. Peter Judge says it might as well measure kittens per kiloWatt hour



    Mark Monroe

    ...

    The Green Grid released a memo from their Global Harmonization Taskforce (GHT) in the middle of March, describing new agreements that the team of global experts had reached over the last 18 months of discussion and negotiation.

    One of the metrics endorsed in the GHT memo of March 14, 2014 is Data Center energy Productivity, abbreviated DCeP.

    DCeP was first described 6 years ago in The Green Grid’s publication, WP#13-A Framework for Data Center Energy Productivity, released in April 2008. The original paper, and the Mar 2014 memo, describes DCeP as a metric that “quantifies the amount of useful work a data center produces relative to the amount of energy it consumes.”



    The paper and memo describe a complicated equation that takes into account the relative value of transactions, the time-based decay of that value, and a normalization factor for the transactions. These last three parameters are set arbitrarily based on each business’ understanding of their IT operations. A business can pick any measure of utility and value for any transaction in their IT infrastructure, and use that to develop a picture of their DCeP value that applies to their business.

    Peter Judge, a London-based IT journalist and consultant, wrote a great piece in Techweek Europe in response to the announcement. In it, Peter wonders if YouTube would be measuring productivity in terms of “Kitten videos per kWh,” which some interpretations of DCeP would lead you to agree. I’d even go so far as to say “Kittens per kWh” might be the right measure of productivity for YouTube!

    The Necessity of Simplicity

    Despite the constant flailing that PUE takes from critics, it is the most used, most effective efficiency metric in the IT industry today. Other contenders from CADE to FVER have not gained the coverage, reporting, or public improvement that PUE has.



    From the first study by LBNL in 2007 to Facebook’s and eBay’s real time, online meters, there has been real change in the industry because of the use of this measure. Since the first reports by Lawrence Berkeley National Labs in 2008 to the real-time Facebook display, reliably reported PUE has dropped from an average of 2.2 to 1.08, a 93 percent reduction in wasted energy.

    I believe one of the biggest reasons for that success has been the simplicity of PUE: hire an intern, send them around the data center with a clipboard, in a couple hours you have an estimated PUE value that a C-level executive can grasp in an instant. One number divided by another, up is bad, down is good, it cost less than $100 to collect your first value. Any metric that requires more from the user is bound to fail wide adoption.

    This is the fundamental problem with DCeP. It’s complicated.

    Sigma Later, Alligator

    First, any equation with a Greek sigma (Σ) in it loses about 80 percent of its potential users. Though it is necessary for the mathematical description of the metric, all those people who hated algebra in the 7th grade will balk at learning the use of this equation.

    Second, because the equation contains arbitrary value constants and functions, it cannot be used to compare one company to another. The Green Grid makes it clear this is not the primary purpose of the metric, but that is what users want the most: to be able to compare their performance against others in their industry, their geography, their business, to see if they are a leader or laggard in providing IT services. The mantra of, “this metric can be used for measuring improvements over time within an organization” doesn’t help; almost everyone who measures performance wants to compare their number with someone else’s.

    Third, even if my organization can put up with the complexity of the math, this will take a long time to implement, collect, manipulate, report, and explain. If one is measuring total data center productivity, and there are 300 applications running in the data center, I have to instrument all 300 apps to collect their transactions, send all the data to a central collection point, process data based on the arbitrary value and utility functions, calculate the time function, sum the normalized transactions over the reporting period, and then figure out how to describe this process and the results to the execs.

    No intern with a clipboard here–collecting and processing this info will require significant programming resources, cost, and time.

    A DCMM-based Alternative

    The Green Grid has been working on their Data Center Maturity Model (DCMM) for a number of years now. Like other maturity models, it provides a road map to improve capabilities in many areas of operations. The Green Grid’s DCMM has 5 levels, spanning the range from “no effort” to “current best practice,” and extending on to “visionary practice: 5 years out” in 46 topic areas dealing with IT and Facilities.

    The chart below shows a suggested change to the DCMM, and a spectrum of performance measurement that can guide organizations from making no estimate of their IT productivity to having a business-based, meaningful measure that can not only tell how much work is being performed by the IT equipment, but can predict what is the maximum capacity of the IT infrastructure.



    By starting with a simple aggregation of raw CPU utilization of some machines in the data center, the company can get an idea of how much resource is being used. By “raw” utilization, I mean at first assume that all machines are equal and a percent of utilization on one machine is the same on another machine. There are plenty of flaws with this plan, but let’s look at the positives:

    • The number can be obtained quickly, at low cost, and unobtrusively in a production environment (back to the intern with the clipboard)
    • Utilization in Level 1 is a percentage, so unit-less. Easy to understand for CxO
    • Using an industry-standard benchmark, like SPECint_rate 2006, converts into work units (SPEC operations)
    • An average utilization figure, and an uncertainty band, can be calculated
    • Total capacity of the IT equipment can be calculated (answer to “how much do we have?”)
    • More analysis can be done by adding adjustments for CPU clock speed and architecture differences
    • Can easily be moved to counting transactions of key business-critical systems



    As an organization matures in its measurement of productivity, more and more sophisticated methods can be used. Accounting for differences between processors, beginning to count critical transactions, developing business-oriented Key Performance Indicators that point out the relationship between IT transactions and revenue or costs are all ways to enhance the performance assessment.

    ACK-poo! Comparing to Others

    Plenty of flaws, yes, but they don’t outweigh the positives of having simple utilization and capacity numbers. Up through Level 2, where industry standard benchmarks are used to convert utilization into operations completed, organizations could begin to compare themselves to others (with the same caveats as early PUE comparisons).

    To me, the best thing that could happen is a bunch of people could start calculating and publishing Aggregated Average CPU Utilization (AACPUU, or ACK-poo!? Bleah. Needs a new acronym). Then analysts can pick it apart, telling why it’s so wrong, how it can be gamed. Then everyone will secretly calculate their value and say “Phew” or “Oh My Goodness!” when they find out how good or bad they are compared to others. Then real change will happen.

    A simple metric like aggregated CPU utilization will show organizations how much capacity they have and how much they are using it. I urge everyone to send an intern into the data center, collect the data on 1000 machines this summer, and see how valuable it is. Make real change in your organization, let the experts argue amongst themselves.

    Mark Monroe is Chief Technology Officer and VP of DLB Associates. He recently served as Executive Director of The Green Grid, an IT industry and end user consortium focused on resource efficient data centers and business computing environments.
    http://www.datacenterknowledge.com/a...cs-productive/

  2. #2
    WHT-BR Top Member
    Data de Ingresso
    Dec 2010
    Posts
    15,049

    There is no magic metric for data center efficiency

    March 24, 2014
    by Peter Judge

    It’s good to see that Facebook has open-sourced its dashboards for data center efficiency, so anyone can show the public how well their data center is doing - but it is happening while the global industry is struggling to find better measurements than the ones Facebook is using.

    PUE (power usage effectiveness) and WUE (water usage effectiveness), as displayed in Facebook’s dashboards, were developed by the Green Grid as simple measures of efficiency. WUE has yet to take off, but PUE has been way too successful.

    Despite the Grid’s original intention that PUE was not to be used as a comparative figure between data centers, that is exactly how it is being used. After all, if companies are pitching for business based on how green their service is, they need a figure to quote, don’t they?

    PUE is being used as a benchmark. But like all benchmarks it has problems, and creates perverse incentives to meet non-real demands. Readers of this blog will know that PUE divides total power by IT power, and doesn’t take any account of how efficient the IT systems thenselves are. So an “efficient” low-PUE data center might be simply idling and completely wasting all its IT power.

    A very good tweak, FVER, has been suggested by British group the BCS. It compares fixed energy use with variable energy use, and encourages data center owners to reduce waste in both IT and the overhead of the center. But I can’t see it ever getting a similar head of steam to that achieved by PUE.

    In any case, both are about making sure that computer cycles are provided with as little energy input as possible, and neither looks at how effectively those cycles are used. What we need is a measure of data center productivity, not data center efficiency.

    Last week, it looks like we might have got just that. DCeP (data center energy productivity), was proposed by a global taskforce, consisting of the Green Grid, along with two US government agencies (the Department of Energy, and the EPA’s Energy Star intiative), Europe’s Joint Research Centre (source of the EU code of conduct) and two Japanese government bodies.

    What is DCeP? It’s very simple - and possibly too simple. It’s the “useful work” done by the data center in a given time, divided by the energy the data center used in that time.

    What does “useful work” mean? The taskforce says “DCeP allows each organization to define ‘useful work’ as is applicable to its business”. So for a retail outfit, the DCeP would measure number of sales per kWh, while for a search engine, it would measure number of searches.

    I assume for Youtube, DCeP is the number of kittens per kWh,and for Facebook, it would be some aggregate of the number of Buzzfeed links, unwelcome adverts and LOLs.

    The taskforce press release says this makes DCeP a “custom and meaningful metric”. In my view it’s more likely to make it a non-metric. There’s no way to compare Amazon’s sales-per-kWh Youtube’s kittens-per-kWh and Facebook’s LOLs-per-kWh. And there’s no way to make day-to-day comparisons in a single company, as the make-up of the business will change, and companies will adjust their own personal DCeP to suit them (for instance, maybe counting the volume of sales rather than the number)

    The taskforce may have simply showed up the futility of the search for a generic measure of data center efficiency. But in the long run that might do us all a favour.
    http://www.greendatacenternews.org/a...ter-efficienc/


    March 26, 2014
    by Peter Judge

    No one really knows how to measure data centre efficiency. There are problems with the Green Grid’s PUE, but the proposed new measure DCeP (data centre energy productivity) looks so vague, it’s hard to imagine anyone using it.

    It’s important to measure how efficient data centres are: electricity is now a major cost, and data centre owners want to use less of it. PUE (power usage effectiveness) simply takes the amount of power a data centre uses, and divides it by the amount that reaches the IT kit. It’s a simple idea and has been very successful – arguably too successful.

    The Green Grid originally said that PUE should not be used as a way to compare data centres, but that is exactly how it is being used. After all, if companies are pitching for business based on how green their service is, they need a figure to quote, don’t they?

    Whose PUE is better?

    PUE is being used as a benchmark. But like all benchmarks it has problems, and creates perverse incentives to meet non-real demands. Because it divides total power by IT power, you can get a great result (close to 1) when the IT kit is burning lots of power, as long as not much is used by other systems such as cooling.

    But what are the IT systems up to? PUE doesn’t care how effective the IT systems themselves are at delivering work, or even processor cycles. So an “efficient” low-PUE data centre might be simply idling and completely wasting all its IT power.

    There is a very good replacement, FVER, suggested by British group the BCS, and developed by Liam Newcombe of Romonet. It compares fixed energy use with variable energy use, so it encourages data centre owners to reduce waste in both IT and the overhead of the centre. Very neat, but to displace PUE would take a huge shift in the data centre community, who have really got behind quoting PUE at ever opportunity.

    In any case, both measures are about making sure that computer cycles are provided with as little energy input as possible, and neither looks at how effectively those cycles are used. What we need is a measure of data centre productivity, not data centre efficiency.

    What about productivity?

    Last week, it looks like we might have got just that. DCeP (data cente energy productivity), was proposed by a global taskforce, consisting of the Green Grid, along with two US government agencies (the Department of Energy, and the EPA’s Energy Star intiative), Europe’s Joint Research Centre (source of the EU code of conduct) and two Japanese government bodies.

    What is DCeP? It’s very simple – and like PUE it’s possibly too simple. It’s the “useful work” done by the data centre in a given time, divided by the energy the data centre used in that time.

    What does “useful work” mean? The taskforce says “DCeP allows each organisation to define ‘useful work’ as is applicable to its business”. So for a retail outfit, the DCeP would measure number of sales per kWh, while for a search engine, it would measure number of searches.

    I assume for Youtube, DCeP is the number of kittens per kWh,and for Facebook, it would be some aggregate of the number of Buzzfeed links, unwelcome adverts and LOLs.

    The taskforce press release says this makes DCeP a “custom and meaningful metric”. In my view it’s more likely to make it a non-metric. There’s no way to compare Amazon’s sales-per-kWh with Youtube’s kittens-per-kWh and Facebook’s LOLs-per-kWh. And there’s no way to make day-to-day comparisons in a single company, as the make-up of the business will change. On top of this, companies will adjust their own personal DCeP to suit them (for instance, maybe counting the volume of sales rather than the number).

    There are hints in the Green Grid’s White Paper that generic DCeP variations might evolve, such as one for e-commerce. At the moment, it’s difficult to say whether the taskforce may has showed up the difficulty of the search for a generic measure of data center efficiency, or demonstrated its futility.
    http://www.techweekeurope.co.uk/comm...kittens-142441

Permissões de Postagem

  • Você não pode iniciar novos tópicos
  • Você não pode enviar respostas
  • Você não pode enviar anexos
  • Você não pode editar suas mensagens
  •