Cloud computing, virtualisation and data centre centralisation are three trends that mean end-users are increasingly removed from the physical hardware that underpins websites and enterprise systems.

Indeed, it is now quite possible to be a sophisticated and heavy user of IT services without having ever seen a physical server.

It is therefore quite easy for end-users to forget – if they ever knew – the degree to which they rely on physical systems in both their work and home lives. Easy, that is, until some data centre outage serves as an unwelcome reminder.

The vulnerability of the digital realm to physical disruption was made glaringly apparent in late October, when ‘super storm’ Hurricane Sandy hit the east of the USA. Heavy flooding affected a number of data centres in New York, taking well- known websites including Huffington Post, Gawker and Buzzfeed offline. 

A common issue during the storm was that flooding interfered with the fuel pumps in backup generators. Employees of hosting provider PEER1 resorted to carrying “half-full 50-gallon barrels” of fuel up 17 flights of stairs to reach the generators on the roof of its data centre.

Meanwhile, it was (dysfunctional) business as usual over the course of the year for organisations who experienced outages that were the result of errors by humans, rather than mother nature.

Other organisations to suffer notable outages during the year include US-based web hosting provider GoDaddy. In September, a major disruption took millions of its customers’ websites offline.

News of the outage escalated after a Twitter user, claiming to represent hacktivist group Anonymous, said that they had launched a DDoS attack against GoDaddy. CEO Scott Wagner quickly denied this claim, saying the cause was rather more mundane.

“The service outage was not caused by external influences. It was not a ‘hack’ and it was not a distributed denial of service attack [DDoS],” Wagner said in a statement. “We have determined that the service outage was due to a series of internal network events that corrupted router data tables.”

And on Christmas Eve, a number of popular sites including video streaming site Netflix were taken offline following a technical fault at cloud provider Amazon Web Services. Amazon.com later apologised and said the fault – which affected its Elastic Load Balancing service – was caused by an engineer deleting key operational data during a routine maintenance process.

Natural disasters aside, outages such as these are being driven by growing technical complexity within the data centre, according to a survey by security and storage provider Symantec. The 2,453 IT professionals surveyed reported that mobile computing (44% of respondents), server virtualisation (43%) and public cloud (41%) topped the table for driving data centre complexity.

Efficient energy

Data centre operators continued to wrestle with the energy efficiency of their facilities. One major bugbear for UK operators has been the CRC Energy Efficiency Scheme, designed to encourage big businesses to cut their carbon emissions by forcing them to report them publicly (among other measures).

The grievance of major data centre operators was articulated by Steven Norris, former transport minister and current president of the Data Centre Alliance, at Information Age’s Future of the Data Centre conference in September. 

He argued that by aggregating their customers’ IT equipment, data centre hosting providers reduce the overall environmental impact of digital business. But a scheme such as CRC punishes them for having high gross energy consumption. 

“On every occasion where we as a data centre industry have taken a company’s data requirements away from an office environment, [the data centre] manages the energy more efficiently,” he said.

When he made his remarks, the CRC scheme was already in question. In March, chancellor George Osborne launched a consultation in relation to the scheme, proposing to simplify the rules in order to cut the cost of compliance, and to remove the performance league table component. In December, after receiving 255 responses broadly in support of its proposals, the government announced that it would go ahead with its amendments to the scheme.

Meanwhile, the industry continued to develop metrics with which to measure and manage data centre energy performance.

In November, the Green Grid – the industry consortium behind the ubiquitous Power Usage Effectiveness (PUE) metric – official endorsed three alternative metrics, having consulted its various practitioner and technology provider members.

Green Energy Coefficient (GEC) is the proportion of a facility’s total energy consumption that is produced by renewable sources. Energy Reuse Factor (ERF) measures the proportion of energy consumed that is exported for reuse outside the data centre, for example by reusing the heat output. And thirdly, Carbon Usage Effectiveness (CUE) evaluates the carbon emissions produced in powering the data centre relative to its IT energy consumption.

The Green Grid says that alternative metrics such as these will help data centre operators build a more accurate picture of the environmental impact of their facilities, and customers to compare potential facilities more effectively.

Space oddity

When it came to data centre occupancy, 2012 began with a bang. According to property advisory CBRE, the first quarter of the year saw an “exceptional” volume of data centre deals in Europe, higher than any quarter in the past two years. 

The company attributed this to pent-up demand. Businesses that had postponed their long-term data centre investment decisions to see what the impact of the euro crisis might be could wait no longer. This uptick was especially pronounced in London, CBRE found. After the first quarter, however, take-up of data centre space returned to 2011 levels – down significantly from pre-credit crunch figures.

Still, the data centre sector is still outperforming most other property markets: “If you look at other real estate segments, such as office space, take-up of data centre space is pretty stellar”, said Martin Carroll, director at CBRE’s data centre practice, in September.

With IT analysts predicting slow growth in IT expenditure in 2013, it is reasonable to expect data centre occupancy to continue to expand modestly during the year. Far harder to predict is what will be going inside that data centre space.