Platform Engineering or DevOps: which one fits your organisation?
Choosing between platform engineering and DevOps? Let's look at how each approach affects developer experience and productivity in software teams.
Today’s software development is defined by a move towards cloud-based technologies and an increasing appetite for distributed development opportunities. Because of this, Platform Engineering has appeared as the next cultural step in the IT industry, coming after the success of DevOps.
The approach is now seen as a competitive must-have: through carefully selected infrastructure automation tools it can raise developer productivity, enable self-service capabilities, and cut down on errors while boosting reliability, in a way similar to Site Reliability Engineering (SRE).
Both Platform Engineering and DevOps are key to delivering software successfully, yet it is vital to grasp how the two approaches differ. That understanding allows you to make well-founded decisions for your organization and reach good software development outcomes.
What is Platform Engineering: a path towards standardized processes and lower cognitive load
In the early days of software development, the developers who wrote the software were also the ones running it. That was a long time ago. Later, as building software in a distributed way gained popularity, Ops and Dev teams separated and worked independently. This was awkward and slow – getting an idea from inception to production could take as long as six months. DevOps was our answer to this. The DevOps philosophy reunited the teams to shorten the gap between changing a system and continuously delivering high-quality software.
When applications are built, tested, released, monitored, and maintained together, code no longer has to be passed from one specialist to another. This is why cross-functional teams with development and operations expertise built in brought major improvements to some software teams. Even so, some organizations did not get the results they hoped for because DevOps roles and responsibilities were defined too vaguely.
As part of adopting DevOps, some organizations moved operational work onto developers, with mixed outcomes. Senior developers can often take care of many duties that ops teams used to handle, but in some cases this piled extra work on them and increased cognitive load, resulting in developer toil. Organizations with a heavy dependence on infrastructure understood that a successful DevOps adoption and operation required offering the right degree of abstraction and structure.
Welcome, Platform Engineer
So, as a developer, you don’t need to stress about mastering Kubernetes, security, or any other DevOps service. Concentrate instead on shipping the features your users need, and leave the remaining areas to other experts. This is where the Platform Engineering concept enters the scene.
A platform engineer builds internal developer platforms that improve the developer experience and raise developer productivity. Platform teams bring all operational services together into a single internal platform that supports continuous delivery, while the people using it – developers, DevOps engineers, SRE engineers – can work at their own pace and rhythm.
What is Internal Developer Platform (IDP)
In the cloud-native era, software engineering keeps getting more complex and demanding, which makes operational excellence more important than ever. Building software today also means building an organization that is agile and autonomous. Platform Engineers create Internal Developer Platforms (IDP) so that software teams can reach dependable tools and workflows and run their applications in production. With these platforms, developers can serve themselves and work efficiently without having to know the whole ecosystem.
Internal Developer Platforms (IDPs) are built to be easy to use and flexible as technologies change. The Platform Engineering team regards every other team as a customer, so IDPs only make sense when several teams need the same services. When just one team needs a given service, that team should build it on its own. Building IDPs aims to establish golden paths for the specialists creating software – that is, to give them clearly defined, dependable methods that make delivering software a better experience.
Spotify was the first company to use standard tooling to enhance its developer experience and boost DevOps productivity.
TL;DR: Spotify’s Platform Developer Experience (PDX) team built CI/CD tools and automated processes after discovering that a DevOps model, in which teams owned their features, caused fragmentation and “development driven by rumors.” To get rid of it, PDX introduced Golden Paths – a collection of best practices for different engineering projects that offers an opinionated, supported way to build services. You can read more on Spotify’s engineering blog.
This illustrates a self-service-driven organization: the platform engineering team gathers the needs of all the other teams and, on that basis, builds an internal developer platform that runs operations for developers, while the platform engineering team keeps improving the platform as requirements evolve.
Characteristics of a good Internal Developer Platform
- It is developed as a product. A product owner makes sure it meets the needs of its clients (developers) well and is pleasant to use (intuitive UI, elegant API, and good CLI).
- It relies on golden paths, so your IDP gives developers standarization, reusability and speed.
- Security and compliance are implemented implicitly.
- Observability is built in.
- It is designed with scalability in mind, so the technology can keep getting better in usability and user experience as needs shift.
Image source: Programming Memes for Developers
Platform engineering teams vs. DevOps teams: how they differ and what they share
Drawing a sharp boundary between these two methodologies is not always easy. Platform engineering is frequently introduced once a DevOps initiative is already in place. Rather than replacing DevOps, it is closer to an evolution of it. The approaches partly overlap, but the two disciplines concentrate on organizing and supporting different stages of software delivery.
For an effective software development lifecycle (SDLC), think about blending different philosophies and borrowing the best practices from several disciplines. Combining DevOps, SRE, Platform Engineering, and other suitable methodologies lets you shape a tailored, efficient, and effective process that fits your specific needs.
No two projects are alike, and each needs its own approach. Building an elaborate development platform may not make sense for a small startup, but depending only on “pure” DevOps can lead to scalability problems for bigger organizations.
Even a large ops team may not be enough to handle every request from developers. What matters most is making developers self-sufficient so they can work efficiently.
In the end, the aim is to give developers a comfortable environment to work in, so they can focus on delivering business value. You have to work out how to find the right balance for your project to succeed.
A DevOps lead at our company
Platform Engineers vs. DevOps engineers vs. SRE
DevOps – short for Development and Operations – is a philosophy. How it is put into practice can differ between organisations. In general, DevOps concentrates on infrastructure that gives dev teams greater control over getting code into production. Day to day, DevOps engineers speed up software delivery and raise its quality while lowering errors and downtime. When the system is available and users are satisfied, you can be confident your DevOps team is doing its job well.
Once code reaches production, SRE engineers step in and concentrate on system reliability. SRE is about keeping the services that software engineers build running and available for potential consumers. At its core, SRE performs work that operations teams have traditionally done, but with engineers who have software expertise. When your system metrics look healthy, your SRE engineers are doing their job well.
Lastly, Platform Engineering grows out of the two approaches above. The concept is shaped around the essential needs of software engineering organisations, especially cloud-native companies. Its focus is on raising developer productivity by making developer self service possible. Successful platform teams design efficient interactions between the teams involved in building distributed systems. The work of platform developers can also be put to good use by Ops and SRE specialists. The faster and more efficiently your various developer teams can work, the better your platform team is performing.
DevOps, platform engineering, or a hybrid?
In modern software development, every development team owns its solutions from start to finish. For some teams, production means end-user applications running in production. For others, it can mean building internal services that other teams consume. You can draw on all of these concepts – DevOps, SRE and Platform Engineering – to build software development processes that are efficient, effective, and scalable.
Each organization’s needs are different, and your primary focus should always be on meaningful collaboration among the specialists who own your software development lifecycle. As software consultants, our approach is grounded in a deep understanding of technology. Working closely with your team, we can learn about your particular challenges and requirements and build practical solutions that raise your developers’ productivity and help you reach your goals.