Rack 'em up: New data centre dynamics

Servers

by Steve Evans| 20 May 2011

Data centres are under more pressure than ever: exploding storage volumes and transaction throughput are pitched against energy, cooling, real estate and budget constraints. Steve Evans considers this tricky state of affairs

Scientists at the University of Southern California recently calculated exactly how much data had been stored across the world. The study took in the years 1986 to 2007, which the university declared the "information revolution" - when people transitioned to a digital world. Covering analogue and digital technologies such as PCs, DVDs, books and floppy discs, the scientists estimated that 295 exabytes of data was stored in the world by 2007.

To put this in context, one exabyte is one billion gigabytes. And that's just up to 2007. The shift to a more digital existence (the study found that 75% of stored information was in an analogue format in 2000, but by 2007 94% of data was stored in digital format) has undoubtedly made this data explosion more pronounced.

Analysts estimate the amount of data being created and stored is doubling every 18 months or so, resulting in a 40-fold increase over the next 10 years.

Add to this concerns over rising energy costs. Gartner reckons that energy-related costs account for approximately 12% of overall data centre expenditure and are the fastest-rising expense. The analyst house also claims that power, cooling, energy supply and cost problems are all likely to worsen during the next few years as organisations emerge from the recession and grow their technology infrastructure, according to a report it released at the end of 2010.

A separate survey carried out by the Carbon Trust found that 49% of the 700 businesses it quizzed are worried about rising energy prices, while a report by the US Department of Energy claimed that less than half the power used by a typical data centre actually powers the IT equipment - the rest goes on infrastructure support such as cooling, power distribution and lighting.

By 2012, the report says, the power costs for the data centre equipment over its useful life will exceed the cost of the original capital investment.

Servers

Wasted pounds
IBM has also carried out its own work in this area. It claims that for many companies electricity is likely to be the second-highest cost to a business after labour. Big Blue also says that for every £1 spent on electricity in an organisation, almost a third - 32p - is wasted.

The Carbon Reduction Commitment (CRC) is a scheme that has the potential to worry businesses. The idea behind it is to encourage UK organisations to reduce their power usage through financial penalties and inducements. It will initially affect 5,000 public- and private-sector organisations that consume more than 6,000MWh of electricity per year, which will typically be those with large UK data centres.

All this points to a huge amount of pressure being placed on data centres, but there is hope. A number of technologies, systems, services and approaches are helping companies get to grips with data centre infrastructure.

Most big vendors have been pushing the idea of a modular or containerised data centre. Also known as a data centre in a box, it offers everything you need to get going: servers, storage, networking equipment and the cooling technology in one package, ready to roll.

The IBM version, called Portable Modular Data Centre (PMDC), can be designed, built, shipped and deployed into any environment anywhere in the world within 12 to 14 weeks. By comparison it can take up to two years for a full bricks and mortar data centre to be designed and built.

"The containerised data centre market continues to develop, with clients asking more knowledgeable questions and identifying better use cases," says Michael Hogan, global offering manager, IBM Site and Facilities Services. "The IBM PMDC alleviates the barriers presented by time and location by offering an immediate, secure and portable solution that answers expanded data centre space and capability requirements more cost-effectively and within a matter of weeks."

HP's offering, called Performance-Optimized Data Centre (POD), and Microsoft's Data Center ITPAC (IT preassembled components) are similar products. Dell, too, is in the game. It offers a modular data centre in a container-style unit, providing either a six-rack or 12-rack configuration. It is also said to be ready to use in around 30 days.

Dell has signalled its intent in this arena by announcing plans to invest $1bn in data centres and other infrastructure over the next year. It will build 10 new data centres across the world, focusing on public and private cloud services and managed IT services, plus helping customers get started in a virtualised world.

Driving down PUE
These investments are not just about building bigger racks with more power and space. There is a genuine effort to reduce emissions and produce more efficient, greener data centres.

Power usage effectiveness (PUE) is a way of measuring energy efficiency. It is calculated by dividing the amount of power entering a data centre by the power used to run it, with a lower score the ideal target. Cisco recently opened a new data centre in Allen, Texas, with a PUE of 1.35, achieved through a number of different features.

The office spaces are powered by solar panels and the facility uses a solid floor and overhead cooling, instead of a raised floor design. Microsoft recently announced an ambitious plan to reduce PUE at all its data centres from 2.0 to 1.25 within two years.

Facebook, which has seen explosive growth over the past few years, has opened a purpose-built facility in Prineville, Oregon. Its initial PUE is just 1.07, and technology in use at the plant means a 38% increase in energy efficiency compared with the commercial co-location data centres it had been using.

"Being able to design more-effective servers, both from a power-efficiency perspective and a cost perspective, is a big part of us being able to build all the stuff we build," says Facebook founder and CEO Mark Zuckerberg.

One of the most impressive new developments in the data centre space over the past few years is that of Thor, which is based in Hafnarfjordur, Iceland, about 15km from Reykjavik. Its mission is to build the greenest data centre in the world. It uses a natural, free cooling system developed in conjunction with Spanish firm AST, and is powered by clean, renewable hydroelectric and geothermal energy sources. It delivers a PUE of around 1.07.

Thor data centre, Iceland
Thor, in Iceland

Analysing the data centre
If you're not too keen on moving to a new data centre, there are approaches to optimise your current infrastructure. Alongside the likes of HP, IBM and BMC, more specialist companies such as Lanamark, Univa, Rackwise and nlyte Software also operate in the market.

IBM has been spending big on analytics over the past few years and has turned its attention to how that technology can be used in data centres. Chris Scott, service product line leader, site and facility services, says: "71% of data centres out there are over seven years old. Effectively, that means nearly three-quarters that exist today are unsuitable for modern, high-density IT deployment."

Why is analytics important to data centre optimisation? Scott points out that it's all about using intelligence to manage change, which sits alongside IBM's Smarter Planet initiative. Analytics in the data centre can be used to determine what the next area of focus should be and enables companies to model outcomes before implementation begins.

Nlyte Software recently unveiled version 6.0 of its data centre performance management platform, which introduced nlyte Analytics to improve the reporting and predicting capabilities of the suite. The dashboard will provide a one-stop shop for all information regarding power, cooling and capacity. The firm claims this can reduce operating costs by up to 20% per year and prolong the life of the data centre by up to five years.

"If you can measure what's going on in the data centre," Rob Neave, co-founder and VP of IT and sustainability at nlyte tells CBR, "you can get the facts and figures about exactly what is going on there. With the predictive capabilities, you can see how much life there is left in the data centre and can possibly hold off buying new equipment."

On the road
Sometimes, however, that just is not possible. There are myriad reasons for making the move, ranging from business growth or merger to space constraints and hitting the limit of your power supply, to reaching server end-of-life and the requirement for higher service levels. If a business has decided the time is right to move to a new data centre - their own or one run by a provider - then what?

"When we talk about data centre migration, we don't just talk about moving a data centre, the equipment and the applications. We're talking about moving an entire business," says Phil Taylor, IBM distinguished engineer. "And the key success criteria is that nobody notices. You don't disturb the business at all and operations were seamless."

As Taylor points out, there is no easy or right way of doing it but there are plenty of wrong ways. So what is IBM's approach?

"We look at it as A, B and A to B. The A is about what compels you to move in the first place and working out what you've got and how to migrate it. The B is the target data centre design - energy and space efficiency, operational TCO efficiencies. The A to B is about the methods and techniques for data centre migration, defining the programme and getting it funded and well structured."

HP, too, offers a step-by-step guide to making to move, called the Five-Phase Transition Approach: it starts with the Migration Start-up Phase, which identifies and assesses the data centre environment; then moves to the Planning Phase; the Verify & Refine Solution, which runs a mock-up of the old environment in the new one, is next; Solution Implementation, which ensures the new infrastructure and application environment continue to function, follows; Close Down & Transition, which transfers all operational responsibilities to new support staff, is the final stage.

A cost case
While all this sounds good, how do you begin to build a business case for data centre migration? "It's very difficult," concedes IBM's Taylor. "How do you make a business case for something that is painful, difficult and high risk? I always think of it as a cost case; you've got to be able to justify spending the money. It could be about risk reduction or regulatory compliance or any number of reasons.

"Tread carefully," concludes Taylor. "These are large, complex programmes. If you try to do it on your own, you're setting yourself up for failure. So choose your partners carefully because you need strong relationships with them. It's a challenging project, but a very rewarding one."

Comments
Post a comment

Comments may be moderated for spam, obscenities or defamation.

Join our network

734 people like this.
0 people follow this.

Servers Intelligence

Buy the latest industry research online today!
See more
Privcy Policy

We have updated our privacy policy. In the latest update it explains what cookies are and how we use them on our site. To learn more about cookies and their benefits, please view our privacy policy. Please be aware that parts of this site will not function correctly if you disable cookies. By continuing to use this site, you consent to our use of cookies in accordance with our privacy policy unless you have disabled them.