Enterprise software development has seen several revolutions, which began with the big bang waterfall approach of project delivery and progressed to agile methodologies. Then came the “shift-left” mantra, whereby developers were given greater and greater responsibility to deliver their code into a production environment, leading to DevOps. These changes in approach have been needed to support developers in a world where there is an ever-increasing workload for digitisation initiatives and digital transformation.
A 2023 Omdia survey of IT operations found that DevOps deployment was fully deployed in less than 25% of organisations. In its Moving to platform engineering doesn’t mean the end of DevOps report, chief analyst Roy Illsey notes that while cost is always going to be a factor, contributing to a failure to deploy DevOps fully, organisational issues and a lack of trust are also barriers.
This, he says, is where platform engineering can help: “DevOps was supposed to be the silver bullet that would solve everything. But nobody really implemented it fully.”
Illsey’s research found people implemented one bit of DevOps, then maybe another bit, and then “got bored”.
Tackling complexity
Much of the complexity in IT infrastructure has come about as organisations shift from application virtualisation to containerisation, Kubernetes and modern cloud-native IT architectures (see box:The shift from virtualisation).
As Csenger Szabo, product manager at Kubermatic, points out in a blog post: “Suddenly, developers were expected to manage everything: infrastructure, deployment, monitoring, security, orchestration, CI/CD [continuous integration/continuous delivery], and more. The burden for teams started increasing, and they started spending more time dealing with the pipeline than actually writing software.”
Platform engineering is the industry’s response to tackling these issues, and was among the topics discussed at the Cloud Native Computing Foundation’s (CNCF) annual KubeCon + CloudNativeCon event in London, which took place in April.
GigaOm field chief technology officer (CTO) Jon Collins says platform engineering is gaining traction as a response to fragmented pipelines, bloated codebases and spiralling costs. But it is not a new concept, and there is no guarantee it will deliver the desired results. “At its core, platform engineering is a way of thinking, reflecting a need to address the realities of software delivery today,” he says.
Platform engineering’s role is to sit between IT operations staff and software developers, where a team provides a platform, which is usually built on top of Kubernetes.
Michael Guarino, the founder of DevOps tools company Plural, says platform engineering on Kubernetes enables faster software releases by providing development teams with self-service tools and automated workflows. It helps to reduce the friction associated with infrastructure provisioning, configuration and deployment, allowing developers to focus on building and delivering new software-powered functionality.
Analyst Gartner forecasts that by 2026, 80% of large software engineering organisations will establish platform engineering teams as internal providers of reusable services, components and tools for application delivery – up from 45% in 2022.
According to Randy Bias, vice-president of open source strategy and technology at Mirantis, platform engineering adds a layer of abstraction and provides developers with a path to production.
But it is complicated. “If you look at the average enterprise, they have no chance in hell of operating like Google. The silos are as siloed as ever, and there are huge barriers between developers and operations people,” says Bias. In his experience, developers don’t want to know about IT infrastructure. “They don’t care about it,” he says. “They want to develop their applications and have them magically deployed.”
If done right, modern platform engineering can deliver “resilience and compliance as-a-platform service”, according to William Rizzo, strategy lead at Mirantis. “Instead of each development team building their own failover logic or wrestling with policy checklists, the platform engineering team provides resilience and compliance out-of-the-box as a menu of composable self-service capabilities,” he says.
Platforms and their component parts need to be efficient, change decisions need to be controlled, and tooling needs to be consistent Jon Collins, GigaOm
Rizzo points out that developers can consume these services quickly, easily and in well-documented ways, with high reliability and low cognitive load.
Platform engineering requires a different mindset
However, as GigaOm’s Collins notes, platform engineering requires a change in mindset. “The mindset needs to be realigned to maintain and improve what’s within the platform, based on need, while still being able to innovate and extend on top of it,” he says.
According to Collins, this approach offers better budget allocation and planning. “Platforms and their component parts need to be efficient, change decisions need to be controlled, and tooling needs to be consistent,” he says.
From an infrastructure perspective, Collins says platform engineering is not about creating wonderfully well-crafted software stacks on top of which everything just works. Rather, it leans on what he calls the façade design pattern – surfacing a subset of functionality that minimises perceived complexity for development teams. For Collins, the platform becomes a product, with release cycles and an assurance of stability.
This concept extends across both newer, cloud-native, and older, virtualised infrastructure. As Collins notes, an application becomes legacy the moment it has been deployed. It’s also a good bet, he says, that decades-old Java and Cobol systems are likely to be around for a few years yet. According to Collins, all such legacy IT systems can benefit from an integration layer, which is a benefit of platform engineering.
The future: AI in platform engineering
While DevOps started with a culture shift and toolset, platform engineering shifts focus, says Chris Aniszczyk, CTO of CNCF. “It’s about better developer experience and streamlining deployment. That means deploying an internal developer platform (IDP), building a curated workflow with smart defaults so developers can spend less time on plumbing and more time delivering value.”
However, Vivek Haldar, vice-president for agents and client innovations at Emergence AI, says that in many companies, adoption of IDP frameworks struggles to rise above 10%. Haldar believes that the primary reason for developer hesitation is that, as with any platform, an IDP has its own learning curve and complexity, and its own challenges at runtime.
Haldar says AI can help overcome this adoption hump. “Millions of developers have ingrained AI into their daily workflow with tools such as GitHub Copilot, Cursor, Claude Code, OpenAI Codex and others. AI copilots and agents have boosted developer productivity and markedly reduced time from idea to implementation,” he says. “Now we can apply this same approach to platform engineering. We can bring the power of copilots and agents to overcome IDP adoption challenges.”
Today’s IDPs are feats of engineering and abstraction. They centralise service catalogues, automate infrastructure and standardise CI/CD pipelines. But he says the developer still needs to understand them to choose the right one for the particular problem they’re solving.
“An IDP still cannot answer the most fundamental developer questions: Why did this break? What’s the right way to do this here? What do I need to ask for?” he says.
Haldar expects that the next generation of IDPs will answer such questions by evolving from a passive menu of options into an active, intelligent partner.
“If a personal copilot in your editor helps you with your code, the platform copilot will help you with everything else, from code structure and design, to deployment, observability and debugging,” he says. “It will allow a developer to state their intent in natural language, and the platform’s job will be to understand that intent and translate it into action.”
