The current response to the topic of green software development is very pleasing. However, this was not always the case. Misunderstandings often arise, especially between practice and research. Does this really have to be the case? As a former researcher, I have changed my perspective to that of a developer and will use examples and some anecdotes to show which misunderstandings occur and how we can counteract them in order to work together towards the same goal.
2. Who Am I?
Head of Technology
@capacura - Europe's most innovative
impact investor
Dr. rer. nat. in the field of modeling and
implementation of methods for the
energy-efficient use of cloud
technologies
Co-development of Blue Angel for
resource-efficient software in
cooperation with the Ökoinstitut and the
Environmental Campus Birkenfeld
10 years of experience in software
development
Several teaching positions at German
universities (of applied science)
Dr. Sandro Kreten
Who Am I?
3. The capacura Portfolio
Founding: 04.2018
Investment Focus: SDG 3, 4, 13*
Number of Startups: 21 Startups
Investment Volume: 13+ Mio.€
Portfolio CAGR: 22,7%/p.a.
capacura Startups
*SDG 3 = good health and well-being; SDG 4 = quality education; SDG 13 = climate action
5. Understanding Green Software
Software Sustainability
Economic
Software
Sustainability
Green
Software
Green
BY
Software
Environmental Sustainability
Human
Software
Sustainability
(Green Software)
IN
Software Sustainability (2001). Calero et al. Springer Nature Switzerland
6. Some Research Results
Criteria for ressource and energy efficient
software
Methodoligies
Measurement Tools
Standard Usage
Scenarios
Recommended
actions for
specific cases
Other Tools
IT
Governance
Human
Factors
ISO/IEC-Norm 14756
Information technology — Measurement and rating
of performance of computer-based software
systems
DIN EN 303470:2018-10
Metrics and Measurements of Servers
ISO/IEC CD 23544
Application Platform Energy Effectiveness
ISO/IEC 33000 Family
Process assessment in the the information
technology domain
9. • Software is complex
• Combined with infrastructure it is a
mess
• Scope often too large to work
comparatively
• “What is the most efficient framework for
machine learning?“
• Finding the “right“ topics is difficult
Complexity
DevOps
Platform Engineer
DevSecOps
CloudOps
etc.
Example
Researcher
10. Lack of hands-on experience
• Methodical competence is available
• Lack of practical experience
• Real life examples often have to be learned and evaluated
• Depth of knowledge and experience of an expert often not
replaceable
Implementation and development of goals often cannot
keep up with the speed of development
11. Great Concepts
• Papers with a good impact factor often require larger concepts
• Practical results need to be justified more often
• Generalization can lead to problems
14. Needs of Target Groups
• Requirements and needs must be specified from the economy
• Quicker results are needed
• Implementation must be cost-effective and result should save costs
• Results must be easy to understand
15. Cooperation, Acceptance and
Exchange of Results
• The acceptance of the content was very difficult until recently
• Open interfaces for monitoring the resources should be made
available to make measurements easier
• Sometimes you need to tweak or hack
tools/frameworks/interpreters in order to find results
• More and easier collaborations are important, which is more than a
hurdle for funding but where it is really about collaboration
16. The good news is…
(even without perfect research results)
17. Results can already be used profitably
• Current research results create a basis for comparative work and for making the
right decisions
• blue angel criteria are a good starting point because of their holistic nature
• Models become more practical and can be applied especially in IT planning
• ISOs
• Especially in the area of cloud and data centers there are helpful insights and tools
• Auto Scalers
• Server Consilidation
• Common processes can already be helpful
• Code Audits
• Refactoring
• Preparing Monitoring and Measurement
• Open source projects to enhance ordinary Monitoring (Code Carbon, RAPL)
• The economy is becoming active. Companies already provide (often open)
solutions
22. A glimpse into the future
• The interest increases
• More developers make results available in the interests of transparency.
Comparisons become easier
• The number of recommendations for action is increasing but they need to be more
practical for developers
• Real Do‘s or Dont‘s are possible but only context sensitive
• Measurement environments become more feasible, less expensive and analysis
becomes automated
• First ML models reveal efficiency gaps
• Although some results already exist, cloud continues to offer very large points of
attack. The scaling effect of savings is simply much greater here.
23. Thank you for the
attention!
I will be happy to answer your questions!