As DevOps adoption rises, software releases hit daily stride

To facilitate faster software releases, DevOps teams now also practice ModelOps, and close to one-third are using artificial intelligence and machine learning algorithms for code review -- more than double last year's number.
Written by Joe McKendrick, Contributing Writer
Two colleagues talking to each other in front of a computer displaying various graphics and code.
Image: Maskot/Getty

DevOps is more than the latest industry buzzword for more agile computing -- it's the only way to get software releases out the door as rapidly as the business demands, while maintaining quality and security. Increasingly, artificial intelligence and machine learning are being brought in to assist in the process. Given the pace of releases, the process is simply too much for anyone to oversee manually. And it's working out well for many IT shops.

That's the word from GitLab, which released a survey of 5,001 technology managers and professionals that finds significant growth in DevOps practices over just the past 12 months. In 2022, a sizeable slice of respondents (47%) indicated DevOps or DevSecOps was their methodology of choice, a five percentage point increase over 2021. 

Also: DevOps Nirvana is still a distant goal for many, survey suggests

With this rise in DevOps comes an increased cadence in software delivery, the survey shows. 7 in 10 DevOps teams (70%) release code continuously -- defined as once a day, or every few days -- up 63% from last year. At least 60% of developers are releasing code faster than before. A full 35% said they are releasing code twice as fast, while 15% are releasing code between three and five times faster. 8% said the code is flying out the door more than five times faster.

To facilitate this, more high-level automation is being applied to software delivery -- the survey finds 62% of DevOps teams are practicing ModelOps, or the governance and lifecycle management of artificial intelligence models. At least 31% of teams are actively using AI and machine learning algorithms for code review, more than double last year's number. The survey also finds 37% of teams use AI/ML in software testing (up from 25%), and a further 20% plan to introduce it this year. Another 19% plan to roll out AI/ML-powered testing in the next two to three years.  

Also: Businesses need DevOps to be competitive but must follow best practices to minimize risk

Paradoxically, the waterfall method of code release -- in which software is designed and then thrown over the wall to QA teams or users -- still prevails in many shops. The percentage of teams using waterfall was up 16% this year over last year, the survey's authors report. "Water/Scrum/Fall" practitioners saw a 23% jump from last year, they add. 

DevOps roles continue to shift, the survey also shows. Developers are taking on ops jobs, ops is cloud or platform-engineering focused, and security pros are more hands-on inside dev teams.  

Toolchain sprawl and security are cited as the most pressing challenges to DevOps-based software deployments, the survey also shows. Toolchain consolidation is a high-priority focus, with 69% of managers or professionals seeking to consolidate their toolchains to address challenges with monitoring, development delays, and negative impact on developer experience. Nearly 40% of developers are spending between one-quarter and one-half of their time on maintaining or integrating complex toolchains -- more than double the percentage from 2021.

Security has surpassed cloud computing as the number one investment area across DevOps teams. However, despite an appetite to shift security left, many companies are still nascent in their approach and results -- only 10% of respondents reported receiving additional budget for security. In addition, 50% of the security professionals in the survey report that developers are failing to identify security issues -- to the tune of 75% of vulnerabilities. 

When asked what they could use to do their jobs better, developers in the study seek more and better code review, automated testing, and better planning (all at 31%). Coming in as a strong second was AI/ML for code writing and review (27%) followed by code reuse (26%). "These responses don't represent any significant deviation from what developers said last year, perhaps underscoring how difficult it is to make systemic process and technology changes," the survey's authors conclude.

Editorial standards