The most common objections to Green Software. And good answers to them. Part 1
Green Software is still new to many people. No wonder that some react sceptically at first. Here, in the first part, I present the first 5 of the top 10 critical reactions and give answers.
1st Software does not consume electricity. The decisive factor is economical hardware.
It’s the other way round: the hardware only consumes electricity because of the software. That is the only reason it is purchased and operated. So green software starts with the cause. Better hardware is important, but can only mitigate the consequences.
2nd Our data centre is already climate-neutral.
Climate-neutral means that data centres buy green electricity and, if necessary, CO2 certificates. Neither of these changes the way electricity is generated in Germany: no less coal is burned, nor is any gas or oil saved. In addition, renewable energies are only available to a limited extent. Therefore, it is all the more important to reduce the electricity demand of data centres through green software. This directly reduces the amount of fossil raw materials that are burned to generate electricity. The motto is: “The greenest electricity is that which does not have to be generated in the first place”.
3rd What is green or not green about this line of code?
Source code looks harmless enough. There is no label on it showing the CO2 emissions. But of course the necessary computing time is a direct consequence of the source code. And a large part of that source code is not written by the authors themselves. Instead, it is integrated in the form of frameworks and libraries, which in turn often do more than is needed. The most important day in a project is therefore probably the day when the software architecture is defined and with it the frameworks.
4th We have always developed in a resource-efficient way. What is new about that?
In many projects, load behaviour and performance play a major role anyway. However, load tests are aimed at maximum situations and performance tests at individual transactions. Neither approach determines the optimal settings for power consumption under normal load. In fact, better performance often requires more resources, and so do higher load limits. Energy efficiency is a different objective, which also leads to different results. This requires dedicated tests that can have quite surprising results. For example, the in principle correct approach of using servers to high capacity does not automatically lead to the most energy-efficient solutions. Green software asks different questions, so you end up with different solutions.
5th Green Software does not make a big difference in power consumption.
There are huge differences between software products that do the same thing. The extreme case is cryptocurrencies. Ethereum claims to have saved over 99.9% of the necessary computing power by switching from Proof of Work to Proof of Stake. But even ordinary software shows big differences: the Federal Environment Agency had various products compared back in 2015. Two word processors differed by a factor of 4, i.e. the same processes consume 4 times as much power when they are executed with a different word processor!
I hope that these arguments will help you in your discussions. Do you have further objections? Feel free to post them in the forum. Then we can exchange views there. Independently of this, the second part of the Top 10 Objections to Green Software will follow shortly.