Are you energy efficient?

Most of today’s organisations are unable to perform their basic business functions without fast and reliable access to data. Indeed, the data centre – the location where this data is stored, secured, and served – has become the engine that powers our economy and our society. However, as our hunger for data grows, so do the requirements of these facilities. These demands have led to concerns about not only the high costs of building and operating data centres, but also their environmental impact.

Data centres consume a tremendous amount of energy to both power and cool the equipment they contain. In fact, they are now estimated to use one to two percent of the entire world’s energy – and this is a rising figure. To date, a number of responsible organisations have put targets in place to reduce energy consumption, doing so due to the well-documented links between the use of energy from non-renewable sources – coal or oil, for example – and climate change. Yet, for many, the importance of energy is often overlooked when deciding on a data centre strategy. This is a short-sighted approach. For one, the availability of energy has a critical impact on reliability and scalability, while its cost forms a key part of the overall operational expenditure.

Metrics for data centre energy efficiency

Energy is a complex and rapidly evolving topic, as the associated cost, availability, and carbon emissions are intensely scrutinised around the world. It is also increasingly the subject of legislation and political debate. As a consequence, organisations need to understand the energy implications of the data centre strategy they choose, to ensure that they’re able to meet their availability, cost, and environmental objectives. Although motivations vary from organisation to organisation, and from stakeholder to stakeholder, most view reducing energy consumption as a positive goal.

PUE is used most often to illustrate the relationship between IT equipment and other forms of infrastructure energy consumption. This approach has many advantages, but also limitations.
PUE can be a useful metric in many scenarios, however, as multiple variables affect it, it’s not easy to make assumptions regarding its potential values. For example, although larger data centres tend to be able to operate more efficiently, it’s not always the case in practice. Additionally, PUE becomes less relevant in certain data centre strategies – in particular, it is a far less relevant metric to users of co-located data centres.

Often, when faced with the complexity of making a decision about the best data centre strategy available to them, many organisations are asking themselves whether or not they even need their own data centres and reviewing options like cloud, co-location, automation and managed services to reduce their data centre footprint. It’s important to note that each of the different location-based strategies – in-house, co-located, and cloud – have a distinctive effect on energy consumption.

Going co-loco

When quantifying the impact on cost, energy consumption, and carbon emissions of all three strategies, several factors should be considered – most notably scale, but also the energy efficiency of the data centre, and the configuration and type of IT equipment it contains. Interestingly, when all of these factors are brought into play, research shows that organisations using a co-located data centre can benefit from lower costs and reduced energy consumption when compared to an in-house data centre.

A co-located data centre also usually produces less carbon emissions, although reporting these emissions may prove a little more complex for the responsible organisation that chooses to do so. The positive effect of this approach is largely due to the outsourcing of numerous energy-related responsibilities to the data centre or cloud computing provider. It’s clear that energy is still being consumed and carbon is still being produced, however, as the providers are servicing multiple clients, they’re able to gain greater efficiencies than most individual clients due to greater economies of scale.

That said, when looking to achieve a near-perfect balance of low cost, energy consumption, and carbon emissions, we typically recommend that organisations use the cloud wherever they can, build only when they must, and co-locate everything else. Adopting a co-located data centre strategy can benefit from lower costs, reduced energy consumption and less carbon emissions when compared to an in-house data centre strategy. This is primarily due to removing the need to invest in power and cooling infrastructure, relying instead on this being delivered as part of the service by the data centre provider. Although the data centre provider may pass some of these costs back as part of the management charge, organisations benefit from the economy of scale afforded by the infrastructure being shared among multiple tenants. For the same reasons, adopting a co-located data centre strategy will usually result in a reduction in the carbon emissions associated with energy usage.

Impact on carbon emissions

These would fall into the ‘GHG protocol’s scope 3 for indirect emissions’, as opposed to ‘scope 2 for purchased electricity’. Reporting on scope 3 is optional, although responsible organisations will often include this within their reporting boundary. Organisations adopting this approach wouldn’t need to report any direct emissions associated with the data centre – known as ‘scope 1 emissions’ –such as those from backup generators or refrigerants within cooling systems.

The difference in cost, energy, and carbon between co-located and in-house data centre strategies decline with scale. Organisations that require larger data centres may therefore decide to build an in-house data centre, as the greater control that this affords may outweigh any benefit in cost, energy, or carbon. Indeed, there are numerous organisations that have built highly efficient data centres with a PUE of less than 1.2, and in these cases it’s very possible that this has brought them greater energy efficiency than with a co-located data centre. These examples tend to be extremely large and significant cost and effort has usually been invested to create such efficiencies.

Again, the use of cloud computing provides even greater efficiencies. Using this strategy, organisations eliminate any direct energy costs and carbon emissions by removing the need to purchase equipment. Although some of these energy costs may be passed back to the client in management fees, cloud computing providers are able to command significant economies of scale and keep these costs to a minimum. Carbon emissions associated with cloud computing are difficult to calculate and likely to fall outside of the reporting boundary for many organisations.

Ultimately, however, evidence shows that these are significantly reduced by the efficiency of this model. For these and many other reasons, cloud computing is an ideal strategy for most organisations. Many recognise that there are still challenges to be overcome before they can move to the cloud completely, and therefore choose instead to migrate individual applications that are well suited to this model, while investing in newer, more efficient and supportive technologies within their existing data centres. For environments where cloud is unsuitable, organisations can gain many benefits from the use of co-located data centres – particularly for smaller environments. For those organisations requiring larger data centres with a high degree of control, they may wish to build their own but they need to understand that significant investments may be required to ensure that these are both economically and environmentally sustainable.