Analyst company Gartner has warned that owners and operators of data centres will face more and more problems associated with power, cooling and space, thanks to the prediction of a significant increase in server sales during the next two years.
Gartner, in a new report “Data Center Power, Cooling and Space: A Worrisome Outlook for the Next Two Years”, believes that this influx of fresh equipment into the data centre is going to trigger a fresh bout of headaches for IT managers.
“While server sales are expected to rise over the next two years, many IT administrators are already grappling with data centre power, cooling and space issues of its current fleet,” said Rakesh Kumar, research vice president at Gartner. “Virtualisation and consolidation projects will help offset some of these issues, but with the snowball effect that these issues tend to create within an organisation, users need to act quickly.”
According to Gartner, the worldwide server market declined by 16.5 percent in revenue and by 16.8 percent in volume in 2009. However, analysts predict that the market will recover from 2010 onward, with a compound annual growth rate (CAGR) of 5.5 percent for shipments and 2.4 percent for revenue from 2010 through 2014.
Gartner’s Kumar states that while servers consume only about 15 percent of the direct energy in data centres, there is a cascade effect. Effiectively, the more servers that exist, the more heat is generated, and therefore, the more cooling equipment is needed.
And to make matters worse, while power, cooling and space did vex many data centre personnel during the economic downturn, the IT market is now on an upswing, and Kumar warns that these issues will become significantly worse as the market takes off.
In an effort to mitigate against this, Gartner advises users to not underestimate the issues, and quantify the problem.
“Users need to get accurate capacity-related data to quantify the impact of infrastructure expansion on the amount of data centre power, cooling and available space. This involves working with facility teams to see what is available and how quickly it could be used up. It also means that the infrastructure and operations (I&O) groups need to work with the application and architecture teams to see what is being planned for the next few years and to translate those needs into energy and space metrics,” Gartner said.
Meanwhile, IT services provider Morse thinks that users should be looking to ‘sweat’ their existing infrastructure first, as “expanding server capacity in the traditional manner is no longer sustainable.”
“With ever-growing numbers of servers consuming more power, expelling more heat and eating up more space, it is clear that simply continuing to expand server capacity in a “traditional” manner is not sustainable,” warned Brian Murray, Principal Consultant at Morse.
“Energy isn’t getting any cheaper, cooling isn’t getting any easier and there is nowhere near an infinite area for data centres to expand into. Instead of simply increasing the number of servers to cope with IT needs, IT departments should first make sure they are using what they already have to its fullest potential before they consider any expansion. Essentially they need to be optimising their data centres.”
Murray thinks that, where possible, IT operations should be consolidated and virtualisation used to minimise the physical infrastructure needed. It cited the example of a leading bank, which by refreshing, virtualising and consolidating its server environment, was able to extend its data centre’s life by at least two to three years.
Even more impressively, power usage was reduced enough to provide £1.5 million of annual savings in energy costs, and a CO2 emissions reduction of 6,500 tonnes per year.
“As well as optimisation, power use needs to be rigorously monitored and controlled, not just in the data centre but across the business, so that IT departments know exactly what their energy requirements are and can predict how any expansion will alter this,” said Murray.
“Yet at the moment 70 percent of organisations don’t know how much power their IT infrastructure actually uses: without addressing this lack of knowledge, they will never address the increased pressure that high-infrastructure deployments will bring.”
Targetting AWS, Microsoft? British competition regulator soon to announce “behavioural” remedies for cloud sector
Move to Elon Musk rival. Former senior executive at X joins Sam Altman's venture formerly…
Bitcoin price rises towards $100,000, amid investor optimism of friendlier US regulatory landscape under Donald…
Judge Kaplan praises former FTX CTO Gary Wang for his co-operation against Sam Bankman-Fried during…
Explore the future of work with the Silicon In Focus Podcast. Discover how AI is…
Executive hits out at the DoJ's “staggering proposal” to force Google to sell off its…
View Comments
when was this research conducted? by whom? where?
first and foremost, "servers" are only half of the data-centre. second: recent (and not so recent) r&d has already born solutions for datacentre 2010, which uses no energy for cooling, which can expand and shrink very dramatically and which spends 50% of energy of datacentre 2008. third: CPU is the key. And CPU makers (Intel, ARM) are advancinn very nicely in making low power, high computing, datacentre possible. Example: XEON 7500. What was 3 racks in 2008 is now one 3U box in 2010. Etc,etc ...
Bad article. Waste of my time.
All servers consume power, but the latest generations of server have much better power management features that the installed base in many data centres. So when you combine this with the increased performance that new Xeon 5600 & 7500 servers offer along with virtualisation, refreshing your installed base of servers can enable you to reduce server count and lower energy bills whilst still meeting the need to deliver increased workload. So, sweating your existing assets may not be the most cost ( or energy ) efficient solution. There's a really useful tool here ( http://www.intel.com/go/xeonestimator ) that allows you to plug in your detail of what servers you currently have and to calculate ROI of replacing with latest generation servers and the OPEX cost savings that this would generate.