Reducing the Cloud Carbon Footprint of bol.

Project Introduction

Under the supervision of a senior engineer, me and a fellow junior engineer took on a project to explore Google’s new Carbon Footprint tool. The tool estimates carbon emissions for cloud projects within the entire organization. Using these estimates, our goal was to see how changes to our cloud configuration would affect emissions.

Given that cloud services can be run in different geographical regions, the carbon impact of that service is tied to the energy profile of that region. To illustrate this, we designed an experiment to determine if running an application in a region powered by greener energy sources would show a noticeable reduction in emissions, as reported by the Carbon Footprint tool.

Methods/Setup

For our experiment, we aimed to select an application that represents a large share of our cloud environment. After thorough analysis, we identified the CloudSQL database as one of the most frequently utilized cloud resources at bol.com.Consequently, we set up a simple application that regularly performs a series of CRUD operations on this database. We then deployed this application in two regions with a significant difference in green energy production levels, while monitoring their emissions in the Carbon Footprint tool over a span of three months.

The Results

The emission data showed a significant overall reduction in emissions of around 25% for the region powered by greener energy sources. This discovery shows the potential of relatively straightforward optimizations. Moreover, it strengthened our belief that it is possible to further reduce emissions by implementing strategies that take environmental factors into account. One such strategy, the “Follow the Sun” approach, involves scheduling cloud resources to operate in regions where it is daytime, ensuring maximum utilization of solar energy. In addition, we envision more advanced strategies that incorporate weather forecast data to predict the availability of solar, wind and hydro energy available in each region.

Some Tips

Our experiment has provided the first tangible evidence that these optimization techniques can deliver substantial environmental benefits. We eagerly presented our results to colleagues across Bol, and their enthusiastic responses have reinforced our commitment sustainability initiatives. In the months ahead we aim to conduct further experiments and explore avenues to raise awareness about Green IT within the entire IT community at bol.com. Our goal is to provide practical solutions for optimizing applications and reducing our carbon footprint.

Thanks for reading!