Cloud Computing in Step with Renewable Energy

When listening to a symphony orchestra, one might get the impression that the many musicians on stage merge into a single unit as soon as the piece begins. In addition to the harmonies, the musicians’ shared dynamics are also striking. All musicians must contribute equally to influence the tempo or volume of the music. This coordination is handled by one person: the conductor.
In a sense, the image of a symphony orchestra can be applied to modern software systems, because distributed system components also perform individual tasks but must be perceived as a single unit. In addition, cloud systems offer flexibility and coordination capabilities. This allows for control over when and where an IT-supported process step should take place. This helps, for example, with load balancing within the system.
If we imagine that we can also integrate sustainability aspects into process control, we arrive at the basic idea of Carbon Aware Computing. In Carbon Aware Computing, the metric of carbon intensity is used to control the operation of an IT system. Sustainability thus becomes the conductor of the system.

How the Carbon Intensity of the Electric Grid Fluctuates

Depending on the time and place at which electrical energy or electricity is consumed, different levels of pollutant emissions are associated with it. This is because not all electricity is generated in the same way. The combustion of fossil fuels to generate electricity produces more emissions than energy generation from renewable sources. In addition to greenhouse gases—which are standardized using the metric of carbon dioxide equivalents—other substances harmful to the environment or human health can also be released into the environment. For the sake of clarity, this blog post focuses on carbon dioxide equivalent emissions and excludes other pollutants.
Carbon intensity describes how many carbon dioxide equivalents are emitted per kilowatt-hour of electricity consumed. Put simply, this metric helps assess the sustainability of an electricity grid. Carbon intensity is always specific to a particular location and time.
The location-specific nature stems from the fact that electricity generation varies by region. As a result, carbon intensity is naturally lower in regions with extensive renewable energy deployment than in regions where fossil fuels are increasingly used to generate electricity. To get an idea of how carbon intensity varies around the world, you can use the online tool Electricity Maps, which provides current and historical data on carbon intensity and electricity generation.
Furthermore, carbon intensity is a time-specific value, as electricity generation is not static. Consumer energy demand and the availability of renewable energy sources are subject to fluctuations. Energy demand rises in the evening, for example, when many people use household and kitchen appliances, while solar energy generation decreases because the sun is no longer shining as intensely. As a result, energy producers continuously generate electricity from fossil fuels to keep the power grid stable. Fossil fuels are used for this purpose because they are dispatchable—meaning the amount of electricity generated can be easily controlled. Consequently, carbon intensity also fluctuates depending on the time of day. An effective measure to counteract these time-dependent fluctuations is the installation of energy storage systems.

What Is Meant by “Carbon-Aware Computing”?

Carbon-Aware Computing addresses this issue. To do so, IT systems are controlled automatically to dynamically adjust energy consumption based on the carbon intensity of the power grid being used. The goal is to reduce carbon-equivalent emissions during software operation. In the field of Carbon-Aware Computing, four patterns have emerged that help manage energy consumption in a sustainable manner:

Location Shifting: A resource is deployed in a geographic region with low carbon intensity to reduce carbon emissions, or a resource is provided in close geographic proximity to the user to shorten the data transmission path.

Temporal Shifting: An IT-supported process is carried out during a period of low carbon intensity in order to reduce carbon emissions.

Peak Shaving: An IT system controls the number of requests processed to avoid peak loads. Peak loads can lead to an exponential increase in energy consumption. This principle can be implemented, for example, by placing requests in a queue.

Demand Shaping: An application reduces its own output when carbon intensity is high in order to lower its own energy consumption.

These four patterns can also be used in combination to maximize the benefits of carbon-aware computing. However, they must be implemented in a way that is consistent with the software’s other quality requirements. For example, the use of location shifting requires that the software be permitted to run in other geographic regions and that security requirements do not prohibit this.

The Impact of Carbon-Aware Computing

• Location shifting and temporal shifting are great ways to reduce CO2 emissions; for example, Poland vs. Scandinavia
• It also increases a system’s resilience, since it does not rely on fixed locations or times and can be used flexibly
• Peak shaving and demand shaping have a positive effect on load behavior and system stability, but may affect the fulfillment of tasks and must therefore be consistent with customer expectations
• Dynamic Pricing: Green electricity is cheaper to generate and could be sold at a lower price
• Relieving the strain on the power grid during periods of high demand caused by other consumers

How Carbon Aware Computing Helps Meet Sustainability Goals

• A great way to make unavoidable resource consumption more sustainable
• It is still important to reduce resource consumption