It’s already a while ago but here some words about my visit to the Cebit in Hannover.  I visited this exhibition with a short interruption of a couple years  since the mid 1990’s. Most of the time I returned home with a freshup of idea’s and the feeling that I got an impression of what’s going on in the digital world. But this year was different. Driving home with my colleague we concluded that this is probably the last Cebit for us to visit. A lot of things we saw looked as old news already discussed in e-magazines and on Linkedin discussion threads. And overall there wasn’t that much to see. Most of the players in the Data Centre world weren’t there. Yes “the system” guys where there with an impressive stand but there connection with the Hannover Messe is more than the just the Cebit. On Data Center’s it was more a German exhibition than a world wide one and even some important German players were missing.  I had the same attitude last week when I started my visit to the Hannover Messe, but this worked out completely different. The Hannover Messe is more a melting pot of all kind of technological ideas. On electrical installations there is a lot of heritage but also quite some innovation. Measuring and controlling is becoming more and more affordable and I have seen a lot of products with which it will much easier to control energy consumption in a Data Centre.  If you don’t know what your PUE or energy consumption is, it got nothing to do with the technological possibilities but everything with the importance of these KPI to you. One other striking observation I would like to mention is the influence of standards.  As soon as we want to communicate we need the standards to state what the capabilities of the different types of equipment are. The IEC 61439 (Low-voltage switchgear and controlgear assemblies)  is a good example of a standard which acts like this. In part 2 different switchgear assembly forms are defined.  All the 4 different assembly forms will give a save installation but only the higher rated assembly types give the possibility for concurrent maintainability and the advantage of  extending a data centre while in operation.  Using these standards in the Program of Demands can greatly help to simplify the definition of your new Data Centre.  Another nice example is the Standard IEC 61000-4-30 (Testing and measurement techniques – Power quality measurement methods).  Power quality analysers complying to class A of this standard can give good information whether the power in your data centre complies with the ITI (CBEMA) Curve. If it does computers should feel happy and no power related problems should occur .  But to analyse the results you must have some good knowledge on power quality in low voltage systems. If you are lost now with all these detailed information I can imagine a bit how you feel. I made a city trip last weekend to Madrid and visited the “Del Prado Museum” after a while it was hard to see the beauty of yet another world famous painting and I got lost too,  until I came in the room with the painting “The Garden of Earthly Delights” by Hieronymous Bosch.

In the very crowded room I stood eye to eye to an incredible visualization of somebodies  view of the world?  500 years ago. There is so much to see in this painting that its hard to explain what you all see but overall it gives you a  complete feeling of a surplus of ideas. This feeling was similar to my feeling of the visit to The Hannover Messe. There are still a lot of new techniques which we are not aware of and which can and shall be used in the data centres over the next coming years.  Data Centres are complex systems. To see what general trends are, you will get more information on visiting dedicated data centre events like datacentre dynamics or Datacentreworld London. If  Cebit wants to keep the data centre public something drastically has to change.

Being first speaker gives the advantage of having a relaxed seminar with a lot of information on different data centre subjects.

The different speakers on data  infrastructure aspects in the DC showed a strong tendency of moving all the cabling to the top of the data centre. Cabling above the racks gives clear vision on what is happening and avoids air path obstruction which might occur with under floor systems.  Fibre optic cabling is winning ground with the ever increasing speeds in the data centre. With either active or passive top of the rack infrastructures. Storage and network cabling topologies are fading into each other with an ever stronger presence of Ethernet in the DC. Intelligent transporting of network traffic between end-user and data centre results in less costs, lower latencies and increasing customer satisfaction. Spreading the tendency of larger data centres, more virtualization  and more use of cloud services.

Back to my own presentation on PUE, here you will find some background information on my Lannews presentation on February  16, 2012 in Brussels. If you would like to receive the complete presentation please send an e-mail to: nvdpas@aegide-group.com.

The Green Grid information can be found in white paper 22 : USAGE AND PUBLIC REPORTING GUIDELINES FOR THE GREEN GRID’S INFRASTRUCTURE METRICS (PUE/DCIE).

In the Netherlands the Dutch Green Building Council starts with  a  BREEAM certification program for data centres, and you can find the next slogan on their site: WORDT UW PROJECT HET EERSTE BREEAM-NL GECERTIFICEERDE DATACENTER?

BREEAM is also active in Belgium but not yet for data centres as I know off.
BREEAM International Assessor Training Course in Belgium look at: http://www.breeam.org/eventdetails.jsp?id=5786

The EU, US and Japan started an initiative to create a common standard for PUE. Resulting in the document : Recommendations for Measuring and Reporting, Overall Data Center Efficiency, Version 2 – Measuring PUE for Data Centers

You will find information on the EU code of conduct for Data Centres here.
For more information on the Code of Conduct you can always contact Minkels.

A lot of  information on PUE you will find in  Our White paper :Tips and Tricks for the professional use of PUE as a management tool.

The ISO-IEC activities can be found here.

25 years ago I visited Venezuela in South America. We entered the country through the Amazon coming from Manaus in Brazil. As we  passed the border the dirt road changed into a paved road and the cars started to grow in size. After a while I asked our Landcruiser driver how the Venezuelans could afford driving such a big cars and he explained that the gasoline was cheaper than drinking water. At first I didn’t believe him but when he filled up his car at the petrol station he had to pay less than a dollar. Just check http://www.marketwatch.com/story/low-gas-prices-plague-venezuela-2011-03-16 and you can read that hardly anything has changed in Cháves country.

Why this story? Well, this is what came up to me when I heard an IT manager talking about energy usage in his data centre. He had no Idea how much energy  his data centre consumed and no clue about PUE. Actually he was not interested in this information because  it wouldn’t make his Data centre more reliable or perform better. Which both isn’t true as I will explain later. But first another short story.

My youngest daughter just became 14 last November and se got a new telephone. As parents  you are uneasy if they bike home alone from school so that why she needed a new smartphone…. . With the new phone came a 250MB data limit. After a couple of days she asked me  how she could get an idea of how much data she had used. I told her that she could wait until the end of the month when the phone bill would be mailed by Vodafone. And what if I use more than 250MB? That will cost you €2.50 per MB I said.  So we ended up looking at internet where  I found a great Android app “Traffic Monitor”. It not only measures the used Internet traffic, but it also tells you how much you use on Wi-Fi and even better it specifies exactly how much every individual running app uses. My daughter found out very quickly that the free downloaded game made contact to the internet every time she played it to download a large amount of data to show an advertisement. She was also very happy to find out that Whatsapp is “good” app with minimal data usage.

This is exactly what is happening in a data centre if you don’t measure where the energy is going you can’t manage it. PUE can be the same for the IT manager as gas mileage for the Venezuelan, if you don’t have to pay for energy you’re not interested in how much you use. And if you want to monitor your PUE because energy consumption is your concern then you need tools to measure it. And the  way to do this correct is to measure all the users off energy and not only the IT and total facility part as is strictly enough to get a PUE figure. Only if you measure all energy using equipment in a DC individual you will get a feeling where the energy savings can be made. While measuring individual equipment we found out that a part of a fresh air system was short circuiting, bringing heated exhaust air straight back in to the air intake. The cooling system was kicking in the compressor resulting in high energy usage while the outside air was supposed to be cool enough to run on free cooling. A outside air deflector solved this issue and dramatically improved the PUE.

And don’t wait until the end of the month or even worse the end of the year to get your kWh and PUE report, but study, as with the phone app, your actual power usage under different circumstances to get a feeling for the performance of your DC.

And what about the IT managers reliability and Performance. Measuring energy to individual equipment gives you almost for free information on electrical currents nearing critical limits. Gives you information on voltage, power factor etc. helping you to prevent or detect critical situations resulting in lower failure rates and higher reliability. No the microprocessor won’t run faster when you measure the power consumed by the computer but if you know how much current is going through the rack PDU it is no risk to install extra cpu power if needed.

I only wonder why Venezuela is not yet crowded with Data centers…  The beaches are nice and the people are friendly. Maybe something for a next challenge?

The software tool Prognose of the British firm Romonet fits perfectly in a comprehensive way of designing data centres. The presentation  showed how it is possible to compare different scenario’s in data centre operation and design , in different climate zones, with different partial loads, with different Tiering levels etc. These simulation result in nice graphs with TCO , PUE and other KPI which are important In the data centre. You can simulate in days what otherwise would costs  weeks to calculate.

On the Minkels stand my British colleagues were very busy in informing customers and showing our solutions.  

Unfortunately I was a little late on the second day for the Google opening session so I dropped in halfway the presentation. Joe Kave gave a very well received insight on impressive below 1.2 PUE  figures of their data centres. Have a look at http://www.google.com/about/datacenters/# with a lot of information on how they achieve this. But keep in mind that your business model is quite different from Google’s. Google controls the whole stack from facility to application. And the control seems easier.

In the panel discussion following the Google presentation Joe Kave stated that it is not always that simple. As soon as the PUE comes close to 1 it is obvious that you have to start working on minimizing the IT energy consumption if you want to have an overall impact on your Data Centre energy consumption. To accomplish this the software guys have to write efficient code, so that the servers, storage and network do the maximum of work with a minimum amount of energy. At Google this is done by making the cost of the service depended on the amount of energy used.  Probably a good way to achieve this result without introducing all kind of software efficiency metrics like the SPECpower_ssj2008.

The DCD was held for the first time in the Excel International convention Centre. The halls are nice with a bit more space than the Lancaster Hotel  but I missed the ambiance off London, the stroll true Hide Park and the feeling to be in a world capital. And organization please don’t serve instant coffee in the afternoon. You don’t have to invite Starbucks but drinkable coffee would be nice.

Down stairs in a large hall Schneider, IBM, Activepower and others showed their new modular concepts. Schneider impressed with a large Uniflair 40 foot cooling container and an equal sized Powercontainer. I missed HP 240a EcoPod(somewhere on a show in Austria) and Schneiders ECObreeze . Colt gave a nice presentation on their modular solution which they build for Verne  in Iceland. All the modules and data centre equipment was pre assembled an pre tested in their factory in the UK and shipped to Iceland. And all this was done within 4 months. So from signing the contract to handing over the key it took 120 days to configure, build, preassemble, ship, assemble and commission Verne Globals 500m² data centre. Hopefully  data will be safer than our savings in Iceland.

http://www.youtube.com/watch?v=BysVEvTJRMc&feature=relmfu

All together there where more than 10 different presentations around all kind off concepts involving modular data centre design.  The form factor hasn’t established yet. For sure its not only a shipping container.  A good observation gives that the 19” inch cabinet is undoubtedly in the heart off every modular design.

Separating hot and cold air is a no brainer but all the answers around fire distinguishing in all those different compartments are not yet given. A nice new challenge. Integration of fire detection with DCIM and BMS  seems obvious, but it introduces more complexity to the operation. I missed the Siemens presentation on this subject. With 5 simultaneous presentations its impossible to attend everything. During some boring presentations it was nice to have the thick conference pack so I could read what I was missing at that moment in the other rooms.

The KBC presentation on their new twin data centre in Hungary was very impressive. KBC is consolidating 23 local server rooms in 4  Central and Easter European countries  to 2 times 2000m² new data centres in one country. A process with all the major components off a successful Project. See: https://multimediafiles.kbcgroup.eu/ng/published/KBCCOM/PDF/COM_RVG_pdf_KBC_start_een_nieuw_computerdatacentrum_op_nabij_Boedapest_EN.pdf

And a lot off presentations on Data Centre Infrastructure Management. The market is moving and becoming more mature. Its not only about how to design a good DC but more and more about how to effectively measure and manage all the processes in the DC.  But take care, implementing a DCIM system can become as complex as starting with a new ERP system. It is probably wise to start simple.

A short 10 minute taxi trip brought us back to London City airport and after a smooth flight to Eindhoven I could just say goodnight before my children went to bed.

It comes up that just identifying the areas where we can improve doesn’t help in getting results. The thing what is missing is a Vision. And that’s the moment where operational excellence comes in. If you want to improve it works much better if you set goals based on  your vision. The Green Grid want data centers to become mature, but like a high school student who wants to enrol on MIT maturity stands for reaching a Goal. If the student doesn’t learn he eventually will become mature but he won’t reach his goal. Look on Wikipedia under operational excellence and you get a first taste what I try to say.

Long time ago I lead the project at Minkels to redesign the sheet metal production facility. In the beginning we were all talking about the kind of machines we would need but didn’t know what to reach until after several long discussions it crystalized to the next goals. We must keep production competitive in a high wage country. With an increasing demand on individual tailored solution for our customers. We translated this in the following strategy.

-the facility must be able to produce unmanned during 8 hours. (overall low staffing highly automated)

-The facility must be optimized for running efficiently for small batches. Ultimately reaching a batch size of 1 product with a setup time per production step of 0 minutes.

Yes there where a lot of sub goals  but we could always relate them to our vision and strategy.

The same is true in the improvement of your data centre. Create a vision on which  your acting is based. Only then it is possible to pick the correct best practices and develop a strategy. Develop the KPI’s you can use to measure your improvement. And yes it is possible like explained in tgg maturity model to gradually improve but in practice this doesn’t always give the best result. If possible make major leaps.  Improving the mileage with my car could be done by changing my driving style but the switch to driving a hybrid car gave the actual change.  Installing a cold corridor will help improving the PUE but building a new DC with fresh air cooling and a large volume approach will give you far more benefits in reaching  operational excellence.  But whatever your goals will be measure your performance otherwise you will never find out if you reached your goals. In the Sheet Metal factory we measure the successful unmanned moments. Being every unmanned shift without a disturbance. PUE can be a good start for your DC. Start simple get an impression of what is realistic goal and go for it.

During the process of adjusting your data centre infrastructure and trying to lower your PUE figure, you will ultimately reach a point where the investments won’t be fully offset by the financial benefits of a targetted PUE result. In other words, there is a break-even point within every achievement towards lowering your PUE figure. A break-even point differs from situation to situation. Basically, there are mainly three parameters with which you can determine your own personal break-even point.

These 3 parameters are:

• level of PUE investment
• benefits of lowering your energy costs
• required payback time

The closer you get to the level of  ”PUE=1.0″, the more effort it will take to further improve your PUE results. They asymptotic effect means that you will eventually come to a break-even point where further investments in PUE improvement won’t pay off anymore. Where the break-even point lays, is different for each and every situation. It also depends on the intended payback time. The longer this scheduled period is, the more investments you are able to make and the more chances you get to lower your PUE any further.

Generally speaking, stretching the payback time a bit will pay of in terms of energy efficiency and energy cost reduction.

*Source: Minkels whitepaper “Tips and Tricks for a professional use of PUE as a management Tool”. In this whitepaper PUE specialist Niek van der Pas, working at the R&D department of Minkels, will share his knowledge. He will point out tips & tricks for the right use of a PUE metric, in order to reach a certain level of operational excellence within your own data centre environment. This whitepaper is meant for being used at the level of corporate server rooms as well as commercial data centre level. You can request a copy of our new whitepaper here and every week we will post a new tip at our blog.

Minkels developed the data centre management system in cooperation with the University of Antwerp and leading customers. The monitoring system comprises all the Key Performance Indicators (KPI’s) required to manage the entire infrastructure of a data centre environment cost-effectively and energy efficient. The system accesses all the real-time data required from the various types of infrastructural equipment in a data centre, including PDUs, chillers, CRAC units, UPSs, coolers, cameras and fire detectors.

Thanks to the open source nature of the platform, it readily enables the integration of existing management software – such as an access control system, a Network Management System (NMS) and a Building Management System (BMS)- in Varicontrol DCIM.

* Source: Minkels whitepaper “Tips and Tricks for a professional use of PUE as a management Tool”. In this whitepaper PUE specialist Niek van der Pas, working at the R&D department of Minkels, will share his knowledge. He will point out tips & tricks for the right use of a PUE metric, in order to reach a certain level of operational excellence within your own data centre environment. This whitepaper is meant for being used at the level of corporate server rooms as well as commercial data centre level. You can request a copy of our new whitepaper here and every week we will post a new tip at our blog.

After that you can go a bit further and opt for an Advanced PUE*. It will ask for some more investments in terms of time and money, but your organization is convinced that PUE knowledge will pay off, so that hurdle is already being taken. The PUE data gathered still is not as detailed as you probably would have wanted, but it will give you enough data to fix some energy inefficiencies and improve your PUE quite a bit.

The final step should be to go for a Master PUE. In this phase your PUE measurement system is measuring on a full continuous base, with a level of granularity built in according to the type of your business. The overall PUE figure and subvided PUE figures for different segments of your systems will give the different levels in your organization specific PUE data they individually need. In this phase, with this information supply, you are able to make targeted modifications and improve your PUE figure to the fullest.

Keep in mind that a PUE-system implementation will cost some time and money in the start-up phase, but the gathering of PUE data can be a fully automated process in the end. So, PUE gathering on a continuous base should finally be an easy task for every organization.

* Source: Minkels whitepaper “Tips and Tricks for a professional use of PUE as a management Tool”.  In this whitepaper PUE specialist Niek van der Pas, working at the R&D department of Minkels, will share his knowledge. He will point out tips & tricks for the right use of a PUE metric, in order to reach a certain level of operational excellence within your own data centre environment. This whitepaper is meant for being used at the level of corporate server rooms as well as commercial data centre level. You can request a copy of our new whitepaper here and every week we will post a new tip at our blog.

Aegide embraces the Code of Conduct initiative by helping to create it, and adopting the best practices in our solutions. Experience how we can help you optimize your datacenter.