What is load average, and why three numbers?
DevOps
Roadmap
I made the plan I wish I'd had when I was getting into DevOps: what to learn, in what order, and which tools to actually use.
DevOps engineer path
The word itself glues two roles together. Dev (developers) write code. Ops (operations, the sysadmins) keep the servers running and make sure everything works. They used to sit in separate teams and argue, and releases kept breaking.
DevOps merged the roles. One person or team owns both the code and how it runs in production: deploy automation, infrastructure, monitoring, the process between teams. That's DevOps.
The operating system that runs the internet. About 95 percent of the world's servers run Linux. It's free, open source, reliable, and can run for years without a reboot.
Work with servers, databases, clouds, Docker, or Kubernetes, and you're working with Linux. Android runs on Linux under the hood too.
A way to pack an app together with everything it needs (code, libraries, settings) into one box. That box (a container) runs on any server and behaves the same way.
Before Docker, this was painful: it works on your laptop, it breaks on the server. A container solves that problem.
Servers, networks, and databases get described in text files. You write "I want 3 servers and a load balancer," and a tool creates all of it in the cloud.
You can rebuild any infrastructure in minutes, store it in Git, and review it like code. Popular tools: Terraform and Ansible.
practice
557 labs with an AI agent on your computer
269 lessons come with labs on Docker and kind. The agent guides you through them like a tutor: asks leading questions, checks your work, and remembers where you left off. It also runs mock interviews with a report on your weak spots.
-
01
Connect the agent
Claude Code, Codex, Cursor, or ChatGPT. One command, then sign in on the site.
-
02
Tell it what you want to learn
Linux from scratch, Docker, Kubernetes, or interview prep. The agent builds the route for you.
-
03
Do the labs on your machine
The agent hands you the files, you fix the broken lab environment, a script checks your work. The solution only comes if you ask for it.
Free and open: the WSL and Linux basics tracks, plus a 5-question mock interview.
why this matters
After the roadmap, you'll be able to answer these
These questions actually come up in interviews. Right now you probably can't answer any of them in full, and that's fine. The idea is to come back to this list in a couple of months and check off what you've covered.
How does the OOM killer work, and why do you need it?
What's the difference between merge and rebase in git?
What is swappiness, and when should you change it?
What's the difference between chroot and namespaces?
Load average is 900, but the load is zero. Is that normal?
Why do servers turn off swap? And when do you keep it on?
What's the difference between a Docker image and a layer?
What namespaces exist in Linux, and what is each one for?
cgroups v1 vs v2: what does v2 add that v1 didn't have?
full deep dive, after you sign in
What happens when you run kubectl apply -f pod.yaml
Twelve steps across four layers: from DNS and TCP to admission, etcd, and kubelet. With real strace, tcpdump, and audit log dumps, plus five branches where the path breaks.
FAQ
A short FAQ before you jump into the roadmap.
DevOps isn't a job title or a set of tools. It's a way to organize work so code gets from a developer's head to production fast and without pain. Deploy automation, infrastructure control, monitoring, processes between teams. All of that is DevOps.
In job postings, a DevOps engineer is the person who does this hands-on: sets up CI/CD, runs Kubernetes, writes Terraform, handles incidents. Related roles are SRE (reliability) and platform engineer (builds internal platforms for other teams). Early on the differences are blurry, and the tools overlap.