Curriculum for

Certified Professional for
Software Architecture (CPSA)^®
Advanced Level:
Module
GREEN

Green Software - Development of resource-efficient applications

Participants are familiar with the global challenges of climate change. They know the increasing energy and resource requirements of IT and their main drivers. They can explain why IT must also address the issue of resource and energy efficiency.

Participants know the possibilities for saving energy and CO₂ through digitalisation. They know the sectors that particularly benefit from digitalisation. The difference between savings in IT and through IT is known.

Participants can quantify the energy consumption of information technology, categorise it in relation to other sectors and name the main drivers. They are familiar with trends and developments in IT and key levers for greater sustainability. They understand how increasing data traffic is linked to the associated environmental impact.

The participants know the fields of action for saving CO₂:

Participants know the regulatory framework and the requirements it contains for saving CO₂. They know the existing certifications, such as the Blue Angel, and can use them appropriately in their work. Participants are familiar with standards such as ESG (Environmental Social Governance) and CSRD (Corporate Sustainability Reporting Directive) and their effects and consequences for companies.

Participants are familiar with the Greenhouse Gas Protocol (GHG Protocol). They are able to explain the scopes and assign emissions to the scopes. They can explain which scopes are affected when operating software in the cloud or on premise. Participants are familiar with ISO 14064-1 as a derivation of the Greenhouse Gas Protocol.

The participants know the motivation for placing green software in the company and the arguments and addressees for successfully anchoring it in the company. They can set sensible targets for the reduction of greenhouse gases for companies. Participants can explain what requirements third parties have along the value chain and how these affect their own company.

Participants know the requirements of the various stakeholders for energy-efficient software and the resulting fields of action. They know the areas with the greatest leverage for reducing greenhouse gases and can prioritize measures accordingly. Participants can classify different software systems and recognize the costs and benefits of optimization.

Participants are familiar with different quality models and can describe quality attributes relating to energy efficiency and categorise them in the quality models. In addition, participants are able to formulate quality scenarios for quality attributes in the area of energy efficiency. Depending on the intended use and the quantity structure of the application, they can consciously decide whether the quality criteria are used for the application or for the optimisation of the development process.

Participants know the positive and negative interactions between quality attributes in the area of energy efficiency and other quality attributes. Based on this knowledge, they can recognise quality requirements that conflict with each other and develop alternative solutions to mitigate conflicts. In particular, the quality attributes of elasticity, scalability, modularity and resilience are discussed in more detail so that participants can assess their influence on energy efficiency.

Participants are able to relate resources to the service provided (business metric). They are familiar with the ISO standard Software Carbon Intensity (SCI) as a metric and understand how it can be used as a business metric.

The participants know the methodological procedure for measuring energy consumption and CO₂ emissions. They understand that it is generally not possible to evaluate/measure absolute consumption very precisely (especially in the cloud), but how to analyse trends instead. They know proxy metrics for CO₂ emissions.

Participants know how to set up meaningful scenarios for measuring energy efficiency.

Participants understand that the procedure for measuring energy efficiency is related to the procedure for measuring performance. Participants understand how to proceed systematically in order to narrow down consumption hotspots.

Participants know different categories of measurement methods and their areas of application. They understand how to combine tools from the different categories for an analysis. Categories include, for example:

Participants are familiar with the challenges of measuring in typical enterprise environments, particularly with regard to distributed components and peripheral systems, hybrid cloud scenarios and compliance requirements. They understand that a differentiated measurement approach is necessary, which varies according to the development phase, environment and objectives.

Participants know the procedures and limits for optimising programs. They can distinguish sensible starting points from less worthwhile goals. Participants are familiar with the energy efficiency of various programming languages and can assess the areas of application for which they are suitable. They can evaluate the runtime behaviour and development speed of programming languages and make the right choice based on the requirements. They also understand the differences between ahead-of-time and just-in-time compilation, bytecode and machine code and the conflicting goals of garbage collectors.

Participants are familiar with the principle of computational complexity and can evaluate and select algorithms in terms of their runtime and energy efficiency. They can use caches in connection with energy efficiency, know how they work (e.g. cache replacement) as well as the advantages and disadvantages. They know the possibilities and limitations of tools for detecting energy code smells.

Participants know methods for handling large amounts of data. They can assess data structures and operations on these data structures in terms of energy efficiency and make the right selection based on the given requirements.

Participants know the pitfalls of using database systems and can avoid them. They are able to evaluate and optimally select different approaches for mapping relationships between entities on the basis of business requirements and quality requirements.

Participants will be familiar with ways in which agentic coding tools can be used to identify energy-related vulnerabilities in code and to suggest or implement improvements – for example, in the choice of algorithms, data structures or resource usage. Participants will be able to use these tools effectively and evaluate their results critically.

Participants can assess different architectural styles such as distributed systems, structured monoliths or serverless and their impact on energy efficiency. In addition, they are able to assess the difference between stateless and stateful components in terms of energy requirements. Participants are familiar with the principles of cloud-native architectures and their impact on energy efficiency.

Participants are familiar with different types of communication (synchronous versus asynchronous) between building blocks in relation to the provision of data, and can assess the formats and protocols used (text-based versus binary protocols) in terms of energy efficiency. They are able to evaluate the influence of data compression on the transmitted data volume and understand that reducing the call frequency and data volume has a positive effect on the environmental balance.

Participants are familiar with different types of database models (relational, NoSQL, etc.) and different database systems (Postgres, DB2, Oracle, etc.) and can make the right choice in terms of energy efficiency. They are able to select a suitable service model (self-hosting versus cloud-based managed service) for their data storage.

Participants are generally familiar with patterns such as those of the Green Software Foundation to improve the energy efficiency of an architecture. They are knowledgeable with the W3C’s Web Sustainability Guidelines on energy efficiency, which must be taken into account in web design. They are familiar with strategies such as browser caching, the use of a CDN and edge caching, and are able to implement these in their applications. They are familiar with different file formats and types of graphic elements (e.g. animations) and are able to assess their energy efficiency.

Participants can assess the energy efficiency of a data centre. They know the Power Usage Effectiveness (PUE) and its advantages and disadvantages.

Participants know the power-load relationship of hardware. They can calculate and estimate the Server Idle Energy Coefficient (SIEC).

Participants are also familiar with other key figures, including the Renewable Energy Factor (REF), Carbon Usage Effectiveness (CUE) and Water Usage Effectiveness (WUE). They can calculate and evaluate these key figures and know how to improve them.

Participants know the main categories of cloud computing services (cloud service models), in particular "Infrastructure as a Service", "Platform as a Service", "Software as a Service" and "Serverless". They can name the main characteristics of these models and weigh up where and how they differ in terms of energy efficiency and CO₂ emission efficiency and where they are similar. In this context, they are aware of the efficiency advantages of containers over virtual machines.

Participants know the different deployment models for cloud environments, in particular "public cloud", "private cloud", "hybrid cloud" and classic on-premise operation. They can demonstrate the opportunities and risks of these variants in terms of energy efficiency and CO₂ emission efficiency. In particular, the energy efficiency of data centres, flexibility in the selection of hardware, overprovisioning and data traffic must be taken into account.

Participants are able to assess ecologically sustainable aspects of the various providers in order to be able to take these into account when making a selection. To this end, they know how to obtain the following information, among others

In addition, participants know the possibilities and limitations of providers when monitoring emissions from their own resource use, as provided by the providers' carbon/environmental dashboards.

Participants are familiar with the challenges of energy-efficient operation and typical anti-patterns such as overprovisioning due to a lack of monitoring, a lack of automation in provisioning or unfavourable geographical distribution.

Participants are familiar with the terms "FinOps" and "GreenOps" and understand how they are related. They know how and to what extent they can influence CO₂ emissions at individual cloud providers by, among other things,

Participants understand the environmental impact of AI systems, particularly generative artificial intelligence with large language models, throughout their entire life cycle. They can distinguish between the phases of data collection, training, deployment, inference and maintenance, and explain their respective energy and resource requirements. They can also apply the terms operational carbon and embodied carbon to AI systems.

Participants are familiar with basic architectural approaches for energy-efficient AI systems, e.g.:

They can compare these approaches in terms of energy efficiency, cost, latency and accuracy.

Participants are familiar with common techniques for reducing the resource requirements of AI models, in particular:

They can evaluate these techniques in terms of training effort, inference costs, model quality and energy consumption.

Participants understand the trade-offs between classic quality attributes (e.g. accuracy, robustness, latency, scalability) and energy efficiency in AI systems. They can make these trade-offs explicit and take them into account in architectural decisions. Participants are familiar with the concepts of green AI and sustainable AI. They understand that efficiency gains can lead to a rebound effect and can identify countermeasures.

Participants are familiar with various strategies of Infrastructure as Code, Continuous Integration and Continuous Deployments. This includes why and when these are usually used, what basic resource requirements they can have and how test execution and code analyses can affect these requirements.

Participants are familiar with strategies for optimising resources and reducing CO₂ emissions from CI/CD processes. Examples of this include the use of peak shaving, time shifting and location shifting.

Participants know how to integrate energy efficiency measurements into the CI/CD pipeline. They understand how (negative) trends can be quickly identified and how data from different sources in the development process can be consolidated.

Participants are familiar with methods for optimising the development and deployment infrastructure. These include, for example:

Participants are familiar with ways to improve the resource efficiency of their test strategies and test environments. This includes, for example, the effect of different test types (load test, system test, integration test, unit test, …​) on resource consumption and the reduction of this through test impact analyses and the demand-oriented, time-limited provision of appropriate test environments.