Decarbonising code by SHIRLEY TARBOTON

For updates and more, join our community 👉 LinkedIn: devoxx-united-kingdom

According to the International Energy Agency, data centres and data transmission networks now account for 2-3% of global electricity use, which is more than the total proportion of energy used by the aviation industry. Bitcoin presently uses around 110 Terawatt Hours each year, or about the yearly energy consumption of small nations like Malaysia or Sweden, according to the Cambridge Centre for Alternative Finance (CCAF). And according to MIT's Technology Review, training and hosting one AI model will produce 50 metric tons of carbon emissions. So, if we - as software engineers - want to play our part in the drive to achieve net zero by 2050, we need to start asking questions about the energy use and emissions of the systems we build, test, and deploy.
In this talk we will consider the impact of the code we write, how we can make code more efficient, reduce the carbon footprint of testing plans, and what factors to review when making decisions around tooling, monitoring and hosting systems. With the prevalence of cloud-based hosting solutions, and the subsequent ease of dynamic scaling, it's easy to overlook the need to optimise code. We're also going to think about what technical architects and app designers can do to drive customer behaviour towards lower carbon outcomes. And we're going to raise big questions that our employers, clients and vendors should consider when they decide to measure carbon emissions and when they select products and services to deliver a low carbon solution.
This talk will highlight some good and bad practises through real life coding examples; and stimulate conversations around carbon and code.