The data centre of the Future

It is frequently said by industry commentators and experts that data centres have not changed a lot at the infrastructure level in the past three decades. Sure, they are bigger, more complex and bristling with more equipment; certainly, they cost a lot more, and do far more work – but they haven’t been subject to the kind of seismic disruptions that have repeatedly disturbed the IT that is housed in the data centre.

Closer analysis suggests this argument is overstated: 451 Research, for example, tracks about 100 suppliers that offer innovations in areas such as modular design, power distribution, management and storage, direct and indirect cooling, advanced optimization, and management software – not to mention innovations that combine and optimize IT at the physical level. Today’s standard racks, blade servers, dual power chains, building management systems (BMS) and cooling systems were little used in the early 1990s.

The data centre industry has itself also changed dramatically in recent years. Global operators such as Equinix or Digital Realty Trust have emerged as part of a now huge global multi-tenant data centre industry, alongside the hyperscale Internet infrastructure data centres such as those run by Amazon, Google and Microsoft. Suppliers such as APC, now Schneider Electric, similarly emerged or matured in the 1990s. Similarly, the adoption of more professional practices and greater standardization in design has led to rising standards. The impact of Tier designs and other practices from The Uptime Institute (an independent division of The 451 Group) and others have contributed to a steady improvement in data centre availability. Energy initiatives introduced by The Green Grid have similarly encouraged greater resource efficiency. But what happens next? Should operators expect any technical or other changes that will render their strategies or their investments obsolete, or at least less economic? Where, if at all, will any big revolutionary or disruptive changes come from in the next five, 10 or 20 years? Will innovations be technological, or perhaps economic and organizational – or both?

There is an argument that, while innovations in IT will continue at a good pace, there will be only small and incremental changes at the physical infrastructure level. Indeed, with many data centres operating at 99.99% availability, and with infrastructural energy efficiency not far from their theoretical best, it be might said that perhaps the two biggest challenges of data centre technology in the past 30 years have been largely solved: availability and efficiency.

451 Research is deeply involved in examining these issues, collaborating with a wide number of organizations and experts. These projects are still underway, and we will be publishing more conclusive findings in the months ahead. Based on this wide variety of work, however, it is possible to see that certain trends are developing and to make some fairly strong arguments. Some of these are obvious, others less so. In this article, we outline why we believe innovation in data centres is not slow and steady, but actually accelerating, and that the resulting business impact will be significant.

In fact, in the next two decades, we expect that most data centres, along with the entire IT infrastructure that is housed within them, will become much more efficient, more agile and capable of achieving higher levels of availability at a lower cost. In part two of this report, The data centre of the future, Part 2: 10 big trends, published in the next issue of Data centre Solutions we outline a list of more specific technologies, practices and drivers that are likely to drive this change in data centres and the data centre industry.

Drivers of innova tion

Effective innovation is never the result of one invention or development, and it never happens in isolation. In data centres, there is a cluster of drivers and developments that are already combining to drive rapid change.

1. Necessity

Even allowing for the ‘steady state’ of innovation in IT, demand for data centre space, power and bandwidth will struggle to keep up with explosive global demand in the next two decades. Although perhaps a little alarmist, some forward thinkers have projected that without major step improvements in efficiency of IT and energy generation and consumption, in 20 years’ time there will not be enough power generation in the world to meet the demand for IT on current trajectories. If data centres were full to capacity, this would be a major challenge. But they are not; in fact, various research studies, supported by some 451 Research initiatives, have concluded most data centres run at very low capacity most of the time – once the IT utilization, and not just power, space and cooling, are taken into account. This presents a huge opportunity for innovation in driving up utilization, moving workloads, and optimizing efficiency and capacity.

2. The power of two (or more)

A second reason – and a common driver of innovation – is that two different areas are converging – IT and infrastructure. (In fact, it could be argued that five areas are converging – processing, storage, networking, infrastructure and industrial control.) This enables more complete approaches to be designed from the outset. Convergence is enabling new and sometimes unexpected expected efficiencies; for example, it is becoming possible to deliver ‘virtual power’ by consolidating and moving workloads using software, thus freeing up resources.

3. Software and data, everywhere

A third factor is the data centres, for all their heavy metal, are increasingly coming under software control, with DCIM (data centre infrastructure management), DCSO (data centre service optimization) and ITSM (IT service management) combining to manage the data centre in real time, autonomically, and using very large amounts of data. This is expected to enable some data centres to ratchet up their efficiency and their availability without substantial infrastructure expense – forcing competitors to respond.

4.Industrial scale

The data centre industry is now not only large, but industrialized – many major infrastructure subsystems, or entire prefabricated modular data centres, are built in factories. Our expectation in the years ahead is that the huge size of the global data centre market, with many operators into a refresh phase, will enable suppliers to produce or preconfigure a wide catalogue of equipment and designs, each optimized for specific situations and applications.

Meanwhile hyperscale operators are innovating in every area to maintain a competitive edge, knowing that at their scale, even small improvements can provide large returns.

5. Cloud drives prices drives innovation.

Over time, many of today’s concerns about cloud computing will evaporate, making price, availability and service differentials between data centres and data centre operators more apparent. This will drive efficiency and innovation, because competing data centres will be under pressure to match prices – and to better understand and reduce their own costs.

As an example of the disruptive impact that cloud services are having, an hour of virtual server time from some cloud providers, with storage and processing, is already less than many data centres are currently paying for an hour’s worth of electricity to power a server and associated infrastructure.

6. There’s more in the labs

There is no law that says that innovation will always continue, or that research and development will always pay. But at present, advances are clear on a number of fronts, spanning IT and software, microelectronics, optics and photonics, energy generation and management, and advanced materials. In the data centre, for example, this covers silicon photonics, flash memory, memristors, power management, fuel cells, prefab data centres, cooling and heat resistance, graphene (for a variety of properties), software-defined data centres and high-speed, high-bandwidth networks. While the exact impact of any one of these can be difficult to predict, some of these technologies, at least, are likely to produce significant changes in economics of running data centres.

The difference between the best-run and the underperforming data centres will widen. This is a significant forecast, because the growing delta between the most and least efficient will cause some operators, and some data centres, to become economically unviable. Fear of ending up on the wrong side of this divide will drive investment and innovation in some – and a decision to outsource in others.