Sustainable, transparent, energy-efficient cloud infrastructures

Open source as a collaboration model

Hi, I’m Tobias Augspurger, open source advocate, engineer and climate scientist. As former head of software development at Deutsche Post/StreetScooter, we focused on open source software for our automotive development. We worked with several companies around the world to implement autonomous driving based on an open source framework called Autoware (1). We were one of the first vehicle manufacturers to decide to build on this emerging framework (2). This allowed us to work cooperatively and modularly with various suppliers and manufacturers around the world on a stack for autonomous driving, which was not common in this industry before. After Deutsche Post decided to stop producing electric vehicles, I was looking for a way to transfer the experience gained from open source deep technology development to the sustainable technology space. Unlike autonomous driving, industry-wide col- laboration was not yet widespread.

I met Tjark Döring, and we decided to work together on open source and sustainable technologies. When the pandemic really took off, we started putting together OpenSustain.tech (3, fig. 1), the directory of open source projects working on sustainable technologies. We received overwhelming feedback and found contributors from all over the world in areas such as renewable energy or sustainable investment.

Fig. 1: On opensustain.tech are all active open source projects gathered that are focused on sustainable technology and energy awareness.

 

The projects are not just simply listed and ordered, but are preceded by an analysis. Is the project active? Is there a community working on the project? Most open source projects become inactive after some time and are no longer developed. Therefore, it is important to keep such directories that you have an overview of what is being further developed. Where can I find contacts and users? Does the community also take care of code quality and documentation?

We were not able to install and test everything in detail, but since it is open source, we were able to quickly get an impression that the projects are being actively developed. It was important for us to see if commits are made regularly, if issues are responded to in a timely manner, and if code quality is high. We performed these analyses for the entire renewable energy and sustainable technology sector and used the results to create Opensustain.tech. Things got difficult when we started looking for projects outside of GitHub. To do this, we talked to many people, asked in communities and forums, and searched for papers through various search engines. It was important that they were true open source projects with repositories that offered the possibility of open participation.

Verifiable sustainability

The creation of the directories has given us a deep insight into the role of open source in sustainable technologies. Therefore, we asked ourselves: What is missing and has not yet been developed? In doing so, we noticed that in the field of renewable energy and especially in energy systems modeling, the #FreetheModels (4) movement is just emerging. They point out discrepancies in the field of renewable energy, hydrogen, electric cars, etc. and open source solutions. Against e.g. political ambiguities, they put fully transparent open source information, models and conclusions and demand the opening of other models as well to be able to establish comparability. In many cases, the public argues based on studies without checking who commissioned the study, how representative or open it is.

This sparked a desire and idea in me to make sustainability verifiable through openness, because sustainability is (still) difficult to measure, industries differ, sustainability approaches vary. But one of our ideas is to change that.

Sustainable open source projects

A precondition for building the open source incubator we called protontypes (5) was that many people had lost their jobs, their fun with open source, and their free time due to lockdowns. The original goal was to create a small incubator where people could creatively develop projects and experiments to see what direction we wanted to go in together. This community was to be open to anyone who was interested. And it’s been growing for about a year now.

Our first project was LibreSelery (6), a new funding model for open source. The experience and contacts we gained here were essential for our following projects. Even though this remained an advanced prototype, we were able to demonstrate a completely new way of funding open source projects and their dependencies. We offered LibreSelery as a product in beta phase.

Unfortunately, GitHub Sponsors, a proprietary funding channel controlled only by Microsoft, also appeared at that time. This funding model is directly integrated with GitHub and was able to displace many other models like LibreSelery very quickly. It has become clear that the time for an open and free funding model for open source has not yet come and that many have not yet realized the danger posed by centralized funding from one of the biggest historical enemies of open source. Since this is a rather radical funding concept, we have abandoned it for the time being and focused on opensustain.tech.

Fig. 2: The electricity map shows the carbon emission on the whole planet (here Europe).

protontypes´projects

After launching OpenSustain.tech, we tried to recombine the projects listed there to explore and show the potential of open source packages and how they can be combined. One of the simplest and most promising opportunities we saw was to identify and automatically offset the CO2 emissions of computing infrastructure by planting trees.

The open-source Scaphandre project (7) was the first building block we needed. It allowed us to determine the energy consumption of Linux machines directly in the software by a hardware abstraction layer. Based on this, we first measure the energy consumption in data centers, then calculate what the carbon footprint is and provide compensation strategies. We know that offsetting emissions is generally a problem because it is usually not verifiable. There are already many projects that only offer offsetting, but it is hard to track how and if the climate gases are really compensated.

However, the first step is to measure the actual carbon footprint and analyze the energy consumption. With knowledge of one’s own energy consumption, one’s own carbon footprint can be calculated and possibly reduced using projects such as electricityMap by shifting high energy loads to areas or time periods of low carbon intensity (8, fig. 2). As long as one’s own processes cannot be changed, the only option is to compensate for the footprint of other projects. To this end, we have initiated the Continuous Reforestation (9) project with Digital Humani (10), which allows us to order the planting of trees worldwide via a simple script. Another open source project helps us to convert the CO2 of a growing tree, so that it becomes clear how many trees have to be planted and maintained in which time period to compensate the carbon emissions. This compensation is an important step, but it must be clear that it cannot stop there. Developing and using renewable energy and reducing energy consumption will lead to true sustainability without the need for carbon offsets (fig. 3).

Fig. 3: The project https://github.com/forc-db/ForC calculate how much CO2 is stored in which forest. If you want to calculate a carbon offset, you need that. Trees at the equator usually absorb more CO2. Even if you plant new trees, you can calculate how much CO2 the forest can store.

 

The second step is to develop a long-term strategy for how the data center can sustainably reduce its climate impact without using carbon offsets. This involves exploring how to integrate renewable energy and reduce the carbon intensity of digital products. Also approaches such as Google’s Carbon Intelligent Computing (11) are needed. It is a way for providers/users to manage computing loads so that consumption is higher when there is a lot of renewable energy on the grid, i.e., there is a surplus on the grid.

We would like to point out for customers which projects and products can be usefully used in combination, e.g. Scaphandre. From the open source projects and tools we analyze, we select those that are necessary in the specific customer environment to analyze energy consumption, look for potential savings, and implement an energy-efficient strategy. In this way, we help companies achieve real, verifiable sustainability that is open and reproducible for others. For us, it is still open whether we will also develop software to adapt the tools to each customer’s situation, or whether we will actually remain only at the consulting level. But that will certainly come up.

Our goal is to get transparent data from which open conclusions can be drawn, this data will be processed with open source tools and projects, the results will be published just as openly, so that a real discourse can also emerge. This is the way to create an open, transparent knowledge process in the field of sustainability.

To this end, we are looking for partners who are willing to disclose their energy data and sustainability approach. We think it helps us a lot when a company discloses this information because it gives them the opportunity to be transparent about how they’ve built their sustainability and how they can improve the results. It also allows them to review and optimize their models in exchange with others. For example, if a company declares its climate neutrality logo and discloses all relevant data, it opens up the possibility of implementing feedback constructively. Such approaches already exist in the field of climate research, where all data is disclosed in a reproducible way to avoid any interpretability.

We have the example of energy consumption in the cloud in mind, how can I measure it and make it more efficient? Something is already happening in this area; Google, for example, has disclosed the data on the CO2 footprint of its data centers (12).

Openness over Sustainability

The next few years will be decisive in determining whether we are able to run our economy based on ecological principles. A massive increase in sustainable inventions is foreseeable worldwide. The success of these inventions will determine whether we can avert foreseeable global catastrophes such as climate change and the loss of natural resources in the long term. If we fail to do this, we run the risk of causing irreversible damage to the entire environment and thus to the basis of life for future generations.

The central question that arises is how to measure sustainable developments per se. How can we ensure that developments have a longterm sustainable impact on our environment? In recent decades, the measurement of corporate sustainability has generated more interest. Several problems have become apparent: lack of mandates and auditing, specious targets, opaque supply chains and complexity. Even with good will, many companies have not succeeded in establishing sustainable processes within their organizations. This has led to sustainable development being outsourced via carbon offset in many cases. The possibility to compensate the impact on the environment by pure indulgence trade has led to a massive price spiral which has resulted in a multitude of dubious projects in this area. In addition, there is a trend to abuse sustainability for pure marketing purposes.

Here another central role of open source and the principles behind it comes to light. Through the use of open tools, data sets and models, it is possible to disclose a company’s own sustainability calculations. This is especially important for investors who are urgently looking for more transparency in the field of sustainable investments. I consider the potential within this area to be so great that it is worthwhile to become entrepreneurial on a larger scale. In the next few months, we will publish our next steps, ideas and projects and are urgently looking for partners on the question: How can openness become the central benchmark for sustainability?

Sources:

  1. https://www.autoware.org/

  2. https://rosindustrial.org/news/2020/5/11/guest-article-on-the-story-of-the-autonomous-logistics

  3. https://opensustain.tech/

  4. https://innovationorigins.com/en/tomorrow-is-good-freethemodels-we-need-open-source-energy-models/

  5. https://protontypes.eu/

  6. https://github.com/protontypes/LibreSelery/https://podcast.sustainoss.org/63

  7. https://github.com/hubblo-org/scaphandre

  8. https://github.com/tmrowco/electricity-map-contrib

  9. https://github.com/protontypes/continuous-reforestation

  10. https://digitalhumani.com/

  11. https://www.gstatic.com/gumdrop/sustainability/24×7-carbon-free-energy-methodologies-metrics.pdf

  12. https://github.com/GoogleCloudPlatform/region-carbon-info

 

Tobias Augspurger,

open source advocate, engineer working in climate science. Software Developer proving sustainability by open source.

Father discovering the beauty of nature with his family.

tobias.augspurger@protontypes.eu