← Journal · · Note
State of DevOps 2024: what's interesting about the cloud
Express42 recently published the results of the State of DevOps 2024 survey, where you can see various trends in DevOps development, the transformation of the specialization, and the use of practices in the cloud. The cloud part was the most interesting for me to watch, since I actively use it in my work.
What stood out:
- The cloud market is slowly growing. There are still concerns about using the cloud due to security and regulatory requirements, but if you need scalable infrastructure and resources on demand, the cloud becomes a great solution.
- About half of respondents use AI assistants in their work (the same ChatGPT). And it really has become good for small tasks like "write a GitLab manifest that does such and such." Personally, I also use it for my research, it helps cut down time spent finding things and structuring information.
- Among managed platforms, databases are used most often. ML platforms and services are still used by less than 10% in the cloud, but I hope that number will grow.
- In terms of skills, DevOps specialists have a growing need for programming language knowledge, since you need to extend your own solutions. It's also still mostly GitLab CI/CD (or similar), Kubernetes, and monitoring.
- In the cloud you also need knowledge of the provider's API, Terraform, Packer, and, surprisingly, Crossplane. I actually want to look into this as an alternative to Terraform, and once I find something interesting I'll write about it.
As a DevOps guy leaning toward MLOps, it's pretty hard for me to understand the boundaries of the technologies I need to have in my toolkit. The T-shape concept dominates a lot right now, where specialists with a fairly broad outlook are valued. For example, my job requires me to have backend skills, data scientist skills, ops engineer skills (and sometimes even product manager, stand-up comedian, and teacher skills). But with the development of AI, access to different specialization areas is getting easier, and now to try something backend-related you don't necessarily need to read a hundred thousand books about Python/Java and understand how they work internally. And the skill of quickly finding information among all this internet noise and applying it correctly, I think that's a very valuable skill right now.
So go check out their report and share your opinion on where the DevOps field will head in the future :)