Enterprises have flocked to adopt platforms that automate and streamline the software development lifecycle, with a huge majority of organizations completing migrations to DevOps or CI/CD platforms during the past two years, according to a recent survey.
In addition to streamlining coding practices, these platforms are also seen as a way to cut back on the number of developer tools in use — and to make the most of AI-powered coding copilots.
Gartner projects the use of DevOps platforms, including GitHub, GitLab, and Harness, will rise to 80% of organizations by 2027, up from 25% just two years ago. In a recently commissioned survey, software delivery solutions provider CloudBees found that 85% of the IT leader respondents had migrated to DevOps platforms in the past two years.
Several factors are driving enterprises the trend, including a desire to modernize app development, cut costs, and reduce the number of apps developers use, says Anuj Kapur, CloudBees’ president and CEO.
“Tool sprawl creates a procurement challenge, having to deal with multiple vendors,” he says. “It creates an integration challenge, having to get tools to talk to each other.”
Many software development leaders also want to eliminate the often-fierce debates within their organizations about the best coding apps and editors, Kapur adds.
“You create these religious wars inside organizations about whose tool is better,” he says. “There’s an effort to address that once and for all with this panacea that becomes, ‘If only we could convert everyone’s religion into one, we’d all be so much better.’”
Cost overruns
But the CloudBees survey comes with a warning: Rip-and-replace migrations, popular for DevOps platforms, come with complications, including unanticipated cost overruns.
Nearly six in 10 of the IT leaders surveyed say their organizations spent over $1 million on DevOps migrations in the past year, with those using rip-and-replace spending an average of $1.75 million. Moreover, organizations that moved to a DevOps platform in a single coordinated event overspent their migration budgets by 18%, more than $300,000 per enterprise. Worse, more than a third of the IT leaders said 25% of that budget turned into sunk costs with no lasting business impact.
As an alternative, more than nine in 10 IT leaders surveyed observe greater efficiencies by integrating coding tools rather than replacing them.
Despite CloudBees being listed as a challenger in the 2024 Gartner Magic Quadrant for DevOps platforms, Kapur recommends that enterprises take the integration approach.
These platforms often offer AI coding assistants, but while there’s been heavy competition among copilots in the past year, much of the generated code has been “so straightforward and arguably so undifferentiated,” he says.
Developers at most organizations prefer a variety of coding tools, Kapur adds. “We have believed from the start that telling developers what tools to use is like telling a teenager what to wear,” he says. “It’s only partially effective.”
Moving to one DevOps platform limits developer choice and locks organizations into one vendor, he adds.
“We don’t believe consolidation is the answer, because in some ways, consolidation goes against the spirit of innovation,” Kapur says. “It basically assumes that a single vendor will be able to address everything.”
Governance vs. flexibility
Warnings about the cost of rip-and-replace migrations resonated with some developers. There’s a real trend toward integrated DevOps solutions, as organizations look for faster delivery, increased security, and easier governance, says Milankumar Rana, architect at software development firm Headstorm.
Unified DevOps platforms streamline governance, auditing, and security, while separate technologies, such as Jenkins, Jira, and SonarQube, provide more flexibility but complicate upkeep, he adds.
For enterprises considering a migration, the cost of retraining developers, creating new continuous integration and delivery pipelines, and improving YAML templates can create budget overruns of 15% to 20%, Rana says.
During Headstorm’s migration between providers, the most significant hidden costs were related to the time spent disassembling systems for compliance evaluations, identity management, permission revisions, and regulatory changes, he adds.
He recommends that organizations take a step-by-step approach to DevOps platform migration. “Begin with non-production pipelines and proof-of-concepts before moving on to critical workloads,” Rana says.
In defense of platforms
While CloudBees’ report raises concerns about migrations, representatives of other DevOps platforms say the problems may be overstated.
The Harness DevOps platform offers tools for continuous delivery and integration, as well as code and security testing, but developers can keep their preferred integrated development environment (IDE), says Jignesh Patel, field CTO at Harness.
“Developers care most about what tools they want to use when it comes to development, but not the tools to do their testing, not the tools to do their security testing, not the tools to do their releases, and not the tools to do their cost optimization,” he says. “They care about their IDEs.”
Patel, who previously served as director of cloud and DevOps at financial services firm Morningstar, directed a rip-and-replace migration to Harness when he was there, and the process went smoothly, he says.
The biggest challenge during a migration is the up-front buy-in, Patel says. “It’s more an organizational than a tooling challenge,” he adds. “Technology is the easiest part of the puzzle. It’s usually the people who are the tougher challenge, and it’s making people believe that this is truly going to add value to their everyday lives.”
Patel also discounts concerns that rip-and-replace migrations can lead to cost overruns. “You’re still letting go of your other tools that you have in your repertoire,” he says. “If you have 15 isolated tools today, once you do rip and replace it with a solution like Harness, you’re not keeping those 15.”
GitHub COO Kyle Daigle also says developers using the platform can keep their preferred IDEs, CI/CD systems, and AI models.
“GitHub has always been built around developer choice,” he says. “We don’t believe in locking developers into a single set of tools nor a single way of working.”
GitHub’s data shows that 90% of businesses rely on open-source code that lives on the platform, Daigle notes. “We believe the real benefit comes from bringing the entire developer lifecycle and the community to the code,” he says. “When you build on GitHub, you aren’t being locked into a platform; you’re plugging directly into the supply chain your software depends on. That proximity is what accelerates innovation.”