While digital usually seems non-physical and ephemeral, digital services run on servers. Servers use energy still mostly produced by burning fossil fuels. This means digital services have carbon emissions and are part of causing climate change.<\/p>\n
Many of the big IT companies are now reporting their carbon emissions. The numbers look similar to the emissions of whole countries.<\/p>\n
In 2018, Amazon emitted 44.4 megatons of carbon<\/a><\/u> \u2013 as much as Finland in that same year. Most of that is third-party carbon emissions, such as from packaging and transport of Amazon goods and emissions from business travel.<\/p>\n Apple\u2019s carbon footprint wasn\u2019t much less in the same year, at 25.2 megatons<\/a><\/u> \u2013\u00a0 similar to Mongolia. If we don\u2019t count emissions from producing Apple-branded goods, their emissions are just 0.6 megatons \u2014 more like the Gambia. Half is from business travel, another 30 percent from employees\u2019 commutes. Google\u2019s carbon footprint was \u2014 1.2 megatons, comparable to Liberia.<\/p>\n Comparing these IT companies\u2019 emissions to countries shows their contribution is significant. So what can you do to minimize your company\u2019s impact?<\/p>\n With the rise of cloud computing, many businesses started migrating their infrastructure to the cloud. The typical approach is renting an always-on server that safely exceeds the business\u2019s current peak load.<\/p>\n But people use the internet, and your services, differently depending on time of day, so loads differ a lot. The load at night is probably significantly lower than at midday.<\/p>\n For software with scalable architecture, load balancers can automatically add more computing power on a single machine or add more servers depending on the number of requests you\u2019re receiving. That means you don\u2019t waste money and energy on servers running in idle mode.<\/p>\n Most cloud providers offer both horizontal (adding more machines) and vertical (adding more power to the machine) load balancing. To use load balancing, you\u2019ll need to build your services with scalability in mind, but once you start, it will save you money and lower your cloud infrastructure\u2019s carbon footprint.<\/p>\n Where does your cloud provider get its electricity? It may depend on the region. Sweden relies mostly on renewable energy, such as wind and water. France heavily employs nuclear power, which is also low emissions. Poland gets most of its energy from burning coal. So, your company\u2019s carbon emissions vary depending on where the data center that provides its cloud computing is based.<\/p>\n Bigger services such as Amazon Web Services (AWS) and Microsoft Azure tend to pay more attention to that aspect of sustainability. AWS users can choose where their computing power comes from using a map that shows which data centers run on green energy and which do not<\/a><\/u>.<\/p>\n Microsoft built a\u00a0low-carbon Kubernetes scheduler\u00a0with a group of scientists<\/a><\/u>. It automatically migrates your tasks to different data centers around the world to maximize green power use, minimizing your carbon footprint. The scheduler can be ported to work with other cloud providers too.<\/p>\n You don\u2019t necessarily need to choose AWS or Azure if you want to go green. The Green Web Foundation has a\u00a0directory\u00a0of smaller providers that use green energy<\/a><\/u>. Running computing on your own hardware usually uses more energy, though. Cloud systems conserve energy by allowing a more even distribution of tasks among computing powers.<\/p>\n Programming languages differ in several ways, including their logic, syntax and capabilities. They also differ in resource consumption<\/a><\/u>. JavaScript and Python tend to consume more resources than compiled programs in languages such as Fortran, C++ and Rust. Object-oriented languages, in turn, consume more than imperative ones. More resources means more power, means more emissions.<\/p>\n Faster\u00a0doesn\u2019t necessarily mean greener. A program may be executed for longer but consume less energy.<\/p>\n You\u2019re more likely to choose programming languages based on the tasks they perform, the abilities of your workforce and the rest of your technological stack, but it\u2019s worth taking into account the energy aspect.<\/p>\n No matter what the language, optimizing code helps lower your carbon footprint, making it faster and less power-hungry. Profiling your software and rewriting ineffective parts can have a significant impact.<\/p>\n The average web page is now more than three megabytes. The bigger the page, the more power needed to transfer it from server to client, and to display it. You may think that doesn\u2019t account for a large part of carbon emissions, but the Green Web Foundation disagrees.<\/p>\n Green Web found running a background video on a Web page resulted in as much carbon as the project team\u2019s commutes. Remember that 30 percent of Apple\u2019s total emissions coming from its workers\u2019 commutes? Now you see the scale.<\/p>\n There are many tools for optimizing web pages.\u00a0Google Lighthouse scores your website on performance, accessibility, best practice and search engine optimization (SEO)<\/a><\/u>. The Green Web Foundation introduced another tool, Greenhouse, that analyzes pages and checks which domains run on renewable energy<\/a><\/u>. It\u2019s not as useful as Lighthouse for optimization, but it can help ecologically savvy organizations choose service providers.<\/p>\n1. Use scalable architecture<\/h2>\n
2. Use greener providers<\/h2>\n
3. Choose your programming language wisely<\/h2>\n
4. Optimize your web pages<\/h2>\n