Planting more than trees: What more sustainable software development might look like

The share of emissions due to digitalization is increasing. Even in 2018, in Frankfurt data centers accounted for 1.3 terawatt hours, or approximately 20% of electricity consumption in the city. At the same time, awareness of the significance – and the necessity – of sustainability is also increasing. Frequently, however, “more sustainable” in the digitalization context is understood in the sense of damage control: Eco electricity and CO₂ compensation by planting trees. One example: Cloud computing provider AWS wants to have converted entirely to renewable energies by 2025 and be climate-neutral by 2040. Precisely for companies that consume a lot of energy performing their core activities, for example, by operating big server farms, such a net-zero policy makes good sense. At the same time, one must ask the question: Is being climate-neutral by 2040 really the best that we can do?

Measures such as planting trees make an important contribution to climate protection, but they’re starting very late, namely only once the negative consequence, in this case CO₂ emissions, has already happened. Anyone who meets their own consumption demands with eco electricity and compensates for the emissions generated is essentially assuming that nothing can be done about the scope of energy consumption. But what would happen if we examined the situation earlier and tried to act as energy-efficiently as possible in general?

To assess the sustainability of software, the following factors must be incorporated, for example:

• Energy consumption: How much energy does development, hosting, and execution of the software cost?
• Energy generation: From which sources does the electricity used come from?
• Life cycle: How long can the software be used? Can it be expanded and supported in the long term?

Important is to examine the process end-to-end if possible – from the server to the end customer’s end user device.

If companies use their own data centers and maintain capacities for special events there, for Black Friday or Christmas, for example, this consumes unnecessary energy throughout the year, for the basic consumption of unused servers is relatively high. Unfortunately, this is still common practice. Cloud computing with optimized utilization presents a more efficient alternative to on-premise servers: Thanks to load-independent autoscaling, resources can be allocated or released as needed, in order to save energy (and reduce costs) and reduce CO₂ emissions.

Transmission also offers savings potential: In mobile communications, 4G (LTE) represented enormous energy savings as compared to 3G (UTMS); now 5G is almost as efficient as copper cable. Even for landline transmission, there are significant differences in energy consumption: Here, FFTH is clearly the most efficient access technology, especially as compared to classic ADSL.

If you consider that streaming makes up a majority of worldwide internet traffic, the question of resolution also becomes relevant: For in 4K, transmission requires approximately twice as much energy as in HD; on mobile end user devices, the difference is often not even visually perceptible. This is a simple lever for consciously saving energy (for unfortunately, providers often play back in the maximum possible resolution by default).