DORA State of DevOps: AI and good leadership for more success and satisfaction

Google's DORA team has published a new edition of its annual State of DevOps report. For the first time, the company considers the impact of artificial intelligence on software delivery performance and explores the question of whether developers only feel more productive with AI or whether they actually are.

The report also summarizes other trends in the current DevOps landscape and provides recommendations for optimizing processes and the well-being of developers in companies.

According to the report, companies achieve higher productivity, job satisfaction and team performance when they promote user-centricity and transformative leadership. In doing so, managers build trust with employees and serve as good role models. This also reduces the risk of developer burnout. Stable priorities are also essential: Frequent changes of direction reduce productivity and increase the perceived stress of those affected.

According to the report, important levers for increasing productivity and flexibility lie in platform engineering and cloud migration. However, companies should approach such initiatives slowly, as cloud usage often involves adapting processes, and platform engineering can also lead to temporary losses in software delivery performance.

In previous years, site reliability engineering or cloud and open source were more important.

The best-performing teams lead in the metrics for change turnaround time, deployment frequency, error rate and recovery time after failed deployments.

Artificial intelligence is also making itself felt in DevOps

The increasing spread of artificial intelligence is also making its mark on DevOps and delivery. Over 75 percent of respondents already use AI for at least one daily professional task, particularly for writing code, summarizing information and explaining code.

Here, the report observes that increased AI usage has led to improvements in documentation quality (+7.5 percent), code quality (+3.4 percent) and code review (+3.1 percent).

In general, the report concludes that AI is a positive factor for productivity, workflow, job satisfaction, code quality and documentation as well as team and company performance. However, the data also suggests that improving the development process does not automatically improve software delivery. Using AI without robust testing mechanisms and incremental updates can reduce delivery by up to 1.5 percent and stability by up to 7.2 percent.

The assessment that AI is not a panacea is also shared by many of the software developers surveyed. Despite measurable increases in productivity, up to 39% of respondents have little to no confidence in AI-generated code.

Heise Conference: CLC 2024

In November 2024, iX and dpunkt.verlag will host the CLC conference – Continuous Lifecycle/ContainerConf – at the Congress Center Rosengarten in Mannheim. Every year since 2014, the event has addressed the most important issues relating to Continuous Integration (CI), Continuous Delivery (CD), Dev(Sec)Ops and GitOps in order to provide answers, expertise and support for everyday project work. From November 12 to 14, this year's CLC will focus on AI-supported DevOps, security and FinOps as well as sustainability.

Highlights from the program

• The AI-supported developer
• Sustainability in the cloud
• Boosting AI Innovation: How AI platforms help to bring your use cases to the street
• OpenTelemetry: What does my distributed system actually do?
• Thinking of a pipeline as a product: modularization, versioning, testing, ...
• The skeleton in the closet: security risks of CI/CD systems
• Workload Identity: The future of secure authentication

Interested parties can still register for the workshops and the conference. Teams or groups of three or more people automatically receive a discount of at least 15 percent in the ticket store.

Google's DevOps Research and Assessment Team (DORA) provides the State of DevOps report free of charge upon registration. It summarizes the responses of over 3,000 participants and takes into account assessments from the Stack Overflow Developer Survey 2023, in which over 90,000 respondents took part, to compare the results.

(mack)