Climate change is the subject of fierce debate and has become one of the defining issues of our generation. As Scientific American reported this month, 2016 was the hottest year in 137 years of record keeping and the third year in a row to take the number one slot, a signal of how much the world has warmed up in the last century. Worries over energy consumption and carbon emissions have become worldwide concerns, with both governments and businesses starting to feel the heat as public demand for more energy efficient practices rises faster than global temperatures.
Once again, it seems the answer may lie in technology. With the transformative power of digital-thinking – including internet, mobile phones, geographic information systems, drones, and remote sensors – it is estimated that we could reduce global CO2 emissions by 20 per cent per year by 2030, the equivalent of 12.1 billion tons of CO2 per year. Impacting a range of different sectors, from agriculture to manufacturing, transport to energy production, eco-driven technology could well hold the key to saving our planet.
The path to energy efficiency, however, is not so simple. With more connected devices in use – estimated by Gartner to hit 25 billion by 2020 – more energy consumption is needed to keep them powered up. Alongside this, vast quantities of energy will be needed to manage, store, process and analyse the 40 zettabytes of data that will be produced by these devices. It is thought that at our current rate, IT energy consumption will end up consuming 10 per cent of the world’s electricity supplies.
This is a serious concern for the IT industry and the need to decouple processing power from energy consumption has already been identified as crucial.
Despite this pressure, IT emissions are expected to decrease to 1.97% of the global total by 2030, from 2.3% in 2020. At the same time, emission reductions attributable to ICT will be nearly ten times greater than energy negative externalities of the ICT sector. More concretely, 1 kWh spent on ICT could save 10 kWh in the economy. This positive trend is the result of innovative start-ups, increased business investment in green-focused R&D programs, as well as several international initiatives such as the Green Grid Sustainable Computing Initiative, the Connect 2020 Agenda and the EU’s energy efficiency directive.
Increasingly, energy performance forms one of the cornerstones of the digital economy – in part, due to the rising costs of running IT systems. The latest generation of supercomputers such as the Bull Sequana, for example, are not only capable of achieving billions of operations per second but are dividing by ten the power consumption compared to previous iterations.
Moreover, the Power Usage Effectiveness (PUE) of datacentres is now almost 1, with many operators working hard to ensure that the energy entering a datacentre is not wasted.
While the hardware side of the industry has made strides in its energy efficiency in servers, datacentres, storage or network for instance, there is more to be done from a software perspective. In fact, despite being ubiquitous in our digitalised environment, the optimization of software is still little explored.
When poorly designed, software impacts the user satisfaction (slowness, productivity, quality of use), resource usage (autonomy, availability) and cost (operation, lifespan of equipment). To combat this, a quality audit is typically carried out to ensure that any code is up to scratch. Results can be significant such as those found on the mobile phone of the French Army which have allowed to multiply by three its autonomy.
Indeed, in addition to detect overconsumption, it may be crucial to predictively assess the autonomy of functionalities in sensitive occupations or for any mobile device with embarked technologies.
However, energy audits are rare, meaning that inefficient processes are often undetected. Given the potential gains, this is surprising.
Additionally, the impact of code on hardware efficiency is still yet to be properly investigated. With developers now using such high-level languages that are abstracted far above the hardware layer, there is a real danger that the code is therefore not optimized for the equipment. Thus, it can cause premature aging, meaning higher costs and poor energy efficiency.
The capability to integrate ecological practices in to the code design will hopefully soon become a leading practice. Algorithm eco-design can have a significant impact on the use of resources, users’ satisfaction and the lifetime of equipment. The resulting gains can increase the autonomy of mobile or any connected devices, deliver significant savings or improve the overall mobile application performance.
Software eco-design is naturally part of the transition to a more circular economy. it helps to optimise the energy use. Considering the virtualization of exchanges and the permanent mobile connectivity, design software “efficient by nature” needs urgently to be considered as a best practice in coding.
Look out for my next blog post, where I will explore the ASCETiC project and Atos’ work in energy conservation…