The data centre: an asset for business growth and agility (International General Counsel, January 2014)

As Europe begins to show the signs of economic recovery post-recession, lowering overheads and improving margins are still very much on the agenda. But a company which approaches a data centre build with the same cost-cutting approach could see its growth stagnate in the long term.

In designing a data centre, CIOs are tasked with the unenviable job of looking after the bottom line in a multi-million pound complex build. Couple this with largely new and complex terms such as big data, cloud computing and virtualisation; new concepts that companies are barely starting to mobilising around. This can stall the willingness to invest beyond a ‘fit for purpose’ only design, but those who do make the leap of faith into these new innovative concepts can find themselves having the competitive edge in a re-growing economy.

Global spending on IT is estimated to be as much as £1.28 trillion in 2014, according to analysts at Worldwide Predictions 2014, fuelled by the rise of mobile data. Due to the exponentially rising demand to cater for mobile and data-hungry customers worldwide, there is a real need for housing information securely where it can be accessed quickly with minimal service disruption.

Data centre growth to sustain this need has been meteoric – investment will reach $148 billion next year according to analysts Gartner. Internet giant Google alone will spend $4 billion annually on its proprietary data systems. The Uptime Institute recently published a report stating that 36% of data centre organisations are receiving large year-over-year budget increases.
Due to the sizes, costs, unique needs plus the advantages that a location can bring, it can be fair to assume that no data centre can be exactly alike; they can be huge installations with dimensions that rival the size of the largest supermarkets, or located over a hundred feet underground under tough limestone bedrock. Some are converted WW2 bunkers, and others can even withstand the blast from a nearby nuclear detonation.

Given that the need for many companies to understandably keep the location and details of their data centres confidential, there is no way confirming how many actually exist, but analysts estimate there are over 500,000 worldwide.

Google opened its latest data centre recently in Taiwan, the first of several in Asia to cater for the booming rise of internet traffic in the Far East continent, home to half of the world’s population. India alone has seen internet users double to 200 million in just two years.

Companies are quickly following suit and co-locating with shared data centre providers or building dedicated data centres in order to offer digital services to new and emerging markets, but it’s quickly turning into a minefield.

We inhabit a world where a customer simply won’t tolerate not being able to log into their account, or that the information they need is offline because a web server is undergoing maintenance. Near-limitless data that is readily available uninterrupted on a 24/7, year-round basis is now the absolute prerequisite, and unfortunately in this era of information on tap, the user sees it as their God-given right.

Data and transaction processing should now be regarded as a second lifeblood of companies that need to be seen as having a consistent online presence. Any disruption to a data centre even for a few seconds causes major damage to not only a firm’s operating ability but their reputation.

RBS has had to deal with a high volume of consumer complaints in the last few years, following several instances where customers haven’t been able to access their savings, leaving many out of pocket, with the blame pointed at the outsourced cost-cutting legacy IT systems built under Fred Goodwin’s reign.

On average, organisations lose around £85,000 for every hour of data centre down time, according to research. Conferring from 2012 sales, internet commerce giant Amazon would be out of pocket £700 for every second its Amazon.com site went down.

In conceptualising and designing a data centre, unfortunately there is no shell scheme (or at least there shouldn’t be) or silver bullet that magically solves a large firm’s IT and data requirements.

Server virtualisation and data centre experts need to work closely with ITOs in order define the specific parameters and requirements a company needs.

Co-location facilities are the industry norm for most mid-size businesses wanting to offload the ‘heavy-lifting’ of data provision onto the experts, with the logistical support taken care of, leaving IT personnel to worry about on-site facilities.

For the large multinationals dealing in information processing and security and performed on a massive global scale could consider investing in their own dedicated data centre, which would make logistical and financial sense given the critical need for speed, responsiveness and importance of their information.

Kerry Partridge, Cisco’s Head of Business Development for Data Centre/Virtualisation, EMEAR, says: “I like to compare the build-your-own data centre versus co-location decision to that of buying a home versus renting an apartment. The home has a greater start-up cost and you're taking on the responsibility to maintain everything, but you're then also allowed to make any changes you want. The apartment (co-location) has little initial cost, your monthly costs are higher, someone else has to maintain it, but you're limited to what changes you can make to the environment. For many businesses it makes great sense to own and operate their own data centre, for others it doesn't. The three key drivers would be cost, security and risk.”

Data centres are also well-known for having gargantuan running costs: the average expenditure per year to operate a large centre is somewhere between £6-15 million depending on the scale of the operation. Yet only around 45% of this goes towards infrastructure: hardware, software, power provision and networking. The majority is spent on heating, cooling, property and labour. Intel suggests that data centres are responsible for 1.5% of global electricity usage.

Power Usage Effectiveness (PUE) has become the industry standard for measuring energy efficiency across data centres. A PUE of 2.0 means for every watt of power dedicated to computing, an additional watt is used for cooling and power distribution. Therefore, a PUE of 1.0 is the end goal in assessing a datacentre for 100% optimum efficiency.

There are several methods of mitigating an unnecessary drain on PUE and hence help ensure a cost-effective strategy. Site selection perhaps plays the most important role – physical proximity to your markets is a good way of helping to ensure fast, reliable data connections, but prohibitive real estate costs might be an issue within large metropolitan areas. Meanwhile, access to people with the right skills to maintain data centres is another factor. Data centres have begun to proliferate in areas like Scotland, Iceland and the Nordics, with colder climates which help lower energy consumption in the server cooling process.

Other businesses might need to take into account legal perspectives such as the storage of data. Partridge says: “Certain countries or industries can have legal restrictions that data must be held in a certain country or location. However, it is possible for the applications to be offshored, so the physical data centre and the applications are run in two different locations. That said, local service may be required for latency-sensitive apps. For most enterprises, the location of data centres is determined by corporate cost, convenience and reliability. No site is ‘perfect’, so you need to prioritise based upon the needs of your project.”

Green solutions in the build of a data centre can help to lower running costs, which may also be driven by environmental legislation and corporate social responsibility. Adherence to the Leadership in Energy and Environmental Design principles look to self-sustaining options in the building, such as the use of insulation schemes, outside ambient light or using free-air cooling over mechanical where possible.

Virtualisation will play a key role in data centres according to a report by Market Pulse surveying IT executives, 79% of EMEA and 90% of US respondents said it was the most disruptive technology in IT strategy. On a basic level, virtualisation refers to the methodology of separating an operating system and its resources from its hardware, hence allowing for multiple systems to run on a single computer. Software as a Service frameworks typically come from virtualisation models. Multiple benefits include reduced power, cooling and hardware costs, and more flexible testing and disaster recovery.

Partridge says: “Virtualisation today is reducing Cisco’s space requirements by a factor of four and power requirements by a factor of two. Virtualisation also avoids using dedicated equipment for individual tasks – automation is made easy and resources can be shared within the data centre more effectively; the advantages of this are flexibility and cost control. Unified data centres bring together the best of server, storage and network as a fully virtualized integrated environment, and a good example of this would be Cisco’s Unified Compute System.”

During the framework outlining process, in order to minimise the aforementioned outages, the options to protect the systems and place contingencies need considering, in order to preserve data reliability and physical security.

“To control costs, you can’t afford much duplication of resources, so this really comes back to the value and importance of the application,” says Partridge. “There are three elements that a company should consider when assessing what measures to put in place. Firstly, what should we protect; secondly, the costs that we are willing to put against this. And thirdly, what is the Recovery Point Objective – how quickly do we need it back should it go down? For example, a bank would place a very different emphasis on their corporate website when compared to their ATM systems across these three categories.”

In this era of big data, mobile and wearable devices and cloud computing, the rationale behind the conceptualisation of a data centre needs to go above and beyond simple cost-driven goals.

Ultimately, not only should a build be situated in an area where geographical benefits can be taken advantage of, but their design should enable a business to have a competitive edge while also giving ample opportunities for rapid growth and quick response to changes in the marketplace. BDMs should symbiotically work in with CIOs to balance IT time and budget measures with the potential for generating business revenue and driving innovation.

Partridge concludes: “Almost every business now runs on some sort of IT, and the data centre is at the core of this. A data centre handles and transforms data, turning data into actionable business information. Better, faster, more available data centres enable a business to act and react, to take advantage of opportunities or stem losses, faster and better and more reliably; and to make daily decisions more effectively. Take Amazon, they were founded as an online book store and have now used their data centre to grow a huge cloud hosting business.”