Digital technologies are both the key to a more sustainable world, and a threat to sustainability. Digital solutions can massively increase the efficiency with which we use resources, but the more sophisticated those solutions become the more resources they consume as they do their work. Part one of this blog series – There’s a paradox at the heart of Green IT – describes this paradox in more detail, and the digital sobriety approach that seems to offer a workable resolution.
Carbon footprinting is standard practice among organizations committed to sustainability, but current methods are too crude to be useful for digital sobriety efforts. A standard carbon footprint audit looks at an organization’s utilities usage (electricity, gas, water, etc.), plus travel, plus a figure derived from expenditure on infrastructure. In an analysis of this kind, most of the carbon cost of an organization’s IT is just part of electricity usage figure. Beyond requiring users to turn devices off when not in use, it doesn’t give any insights into how to reduce that number.
Using crude carbon footprinting on IT is like trying to reduce greenhouse gas emissions from cars simply by recommending that people drive less – it doesn’t address the central problem, which is the design and function of the cars themselves. Similarly, standard carbon footprinting doesn’t look inside an organization’s IT box, it just recommends using the box less.
Looking inside the IT box….
There are three ways of breaking the IT box down for more useful sustainability insights.
Data center power efficiency
Data centers account for about 45% of IT’s total carbon footprint (up from 33% ten years ago) . Improving the power efficiency of these big contributors is a high priority, and there is a widely accepted metric called Power Usage Effectiveness (PUE) (ISO/IEC 30134-2:2016). The data center PUE average has been trending down for more than a decade, from 2.5. in 2007 to 1.6 today (the theoretical ideal is a PUE of 1.0)
Choosing a data center provider with a low PUE score is a strong and relatively straightforward digital sobriety measure.
Securing more energy efficient servers is good, but using them less often is even better. Several devops measures can help here, most importantly: developing apps that manage peak loads without spinning up extra resources, deploying docker-type technology instead of multiplying complete virtual machines, and ensuring projects have precise and lean requirements. It’s also important to make sure these measures don’t result in compute loads being transferred to the client, which would negate server-side energy savings.
If one aim is to maximize server power efficiency and the second is to improve the efficiency with which applications use those servers, the third should be to reduce the number of applications an organization is using (or, as is sometimes the case, maintaining without using). Application rationalization for digital sobriety employs the same mantra as for cost cutting: eliminate duplication, decommission, and modernize.
It’s no accident that digital sobriety measures closely resemble IT cost-cutting moves. Their goals are two sides of the same coin: reducing inefficient use of IT resources to cut costs also cuts carbon emissions. Except it’s not always that simple. There are cases where lowest cost does not equate with lowest carbon – the most cost-effective server vendor may not have the lowest PUE for example.
Creating a program to optimize your cloud setup from a cost standpoint is complex and best accomplished with a dedicated FINOPS team. It is possible to complement FINOPS objectives, which focus on cost, with GREENOPS, which focuses on energy impact. Good work in GREENOPS will also contribute to a good FINOPS performance. Everyone should be able to benefit from this. I’ll be looking at the specifics of GREENOPS considerations in part three of this series.
Read part 3: How to create a Green IT virtuous circle
Read part 1: Why there’s a paradox at the heart of Green IT