A starting point for anyone new to DevOps. It covers why DevOps is a culture and a methodology rather than a tool, the full flow that carries code to production, a map of the technology from Linux and Docker through CI/CD, Kubernetes, and monitoring, and a step-by-step learning roadmap and practice plan.
Beginner
Updated Oct 1, 2026
~7 min read
12 sections
1 code examples
Learning Paths that include this guide
This guide is one step in the paths below. Study it together with the steps before and after.
Understanding DevOps Intro & Roadmap as a real-world workflow
A starting point for anyone new to DevOps. It covers why DevOps is a culture and a methodology rather than a tool, the full flow that carries code to production, a map of the technology from Linux and Docker through CI/CD, Kubernetes, and monitoring, and a step-by-step learning roadmap and practice plan. Rather than listing concepts, this guide is organized so you can follow the content in the order you have to make decisions on a real project.
Key perspective
Infrastructure / operations
Rather than memorizing install commands, understand the order in which traffic, runtime, observability, and incident response connect.
The big picture of DevOps conceptsUnderstand the deployment pipelineMap of the DevOps tech stackDevOps learning orderExploring a DevOps/SRE career
Architecture diagrams
To make what you read stick, it helps to capture the flow as a picture first. The two diagrams below are reference maps you can keep coming back to while learning DevOps Intro & Roadmap.
Learning flow
Rendering diagramβ¦
Architecture view
Rendering diagramβ¦
What DevOps is: a way of working, not a tool
When you first open DevOps Intro & Roadmap, the big picture matters more than individual commands. This section starts with the problems the concepts ahead were created to solve.
In the past, the development team (Dev) wrote the code and handed it over, and the operations team (Ops) deployed and managed it. Developers wanted to "ship features fast" while operations wanted to "reduce change and stay stable", so the two teams were always in conflict. Deployments happened once every few weeks as an all-night job, and incidents turned into blame games.
DevOps is the culture and methodology of tearing down that wall and deploying "often, small, automatically, and safely". Tools such as Docker, Kubernetes, and CI/CD are the means of reaching that goal.
Aspect
Traditional way
DevOps way
Deployment cadence
Once every few weeks or months
Several times a day
Deployment method
People logging into servers by hand
Automated pipeline
Infrastructure
Servers configured by hand (snowflake servers)
Defined as code (IaC)
Incident response
Found out from user reports
Found first through monitoring and alerts
Responsibility
Dev and Ops separated
The team that builds it runs it (You build it, you run it)
How code reaches production
How code reaches production is a point where the options diverge. Using the table to compare what each approach is for and how it differs in operation makes later decisions much easier.
Every DevOps tool automates one stage of the flow below. Keep this loop in your head and, when you meet a new tool, you can tell right away "which stage it belongs to".
Incident retrospectives, improvements fed back into code
Postmortems
-
DevOps technology map: the order to build it up
DevOps technology map: the order to build it up is a point where the options diverge. Using the table to compare what each approach is for and how it differs in operation makes later decisions much easier.
DevOps technologies form layers that depend on each other. If you learn an upper layer without the one below, you cannot find the cause when something breaks. A large share of Kubernetes incidents turn out to be Linux networking or container configuration problems.
What containers changed is a point where the options diverge. Using the table to compare what each approach is for and how it differs in operation makes later decisions much easier.
A container bundles the application and everything it needs to run (runtime, libraries, configuration) into one image so it runs identically anywhere. Because containers are far lighter and start far faster than virtual machines (VMs), they became the standard unit of deployment in modern DevOps.
Aspect
Virtual machine (VM)
Container
Unit of isolation
A whole OS (including the guest OS)
A process (shares the host kernel)
Size
Several GB
Tens to hundreds of MB
Startup time
Tens of seconds to minutes
About 1 second
Density
Dozens per server
Hundreds per server
Main use
Strong isolation, running a different OS
Microservices, CI/CD, fast scaling
Your first pipeline: test and build an image on push
Here you look at Your first pipeline: test and build an image on push alongside real code. Rather than copying the example as-is, read it with an eye on the inputs, the outputs, and the parts most likely to change.
The GitHub Actions workflow below runs the tests on every push and, when they pass, builds a Docker image. These 20 lines are all there is to CI, and adding a deploy step on top turns it into CD.
DevOps performance metrics: the 4 DORA metrics is a point where the options diverge. Using the table to compare what each approach is for and how it differs in operation makes later decisions much easier.
Whether DevOps is working is judged not by which tools you adopted but by numbers. The four metrics defined by the Google DORA research team are used as the industry standard. The central finding of the research is that speed (the first two) and stability (the last two) do not trade off against each other: teams that are good are good at both.
Metric
Meaning
What a good team looks like
Deployment frequency
How often you deploy to production
Whenever needed (several times a day)
Lead time for changes
Time from commit to running in production
Within a day
Change failure rate
Share of deployments that led to an incident or rollback
Kept low (e.g. 15% or less)
Time to restore
Time from an incident starting to recovery
Within 1 hour
DevOps Β· SRE Β· Platform engineer
DevOps Β· SRE Β· Platform engineer is a point where the options diverge. Using the table to compare what each approach is for and how it differs in operation makes later decisions much easier.
Job titles and scope differ from company to company, but the skills needed mostly overlap. DevOps engineers with backend development experience are valued especially highly because they understand the pain points of development teams.
Model serving, GPU infrastructure, quality monitoring (LLMOps)
Before you start
Before you start is a point where the options diverge. Using the table to compare what each approach is for and how it differs in operation makes later decisions much easier.
For a practice server, Docker Desktop or WSL2 on your PC, or a cloud free tier, is enough. You can start local Kubernetes with kind or minikube.
Area
What
How much
Linux commands
Files, permissions, processes, packages, logs
Required β first of all
Networking basics
IP, ports, DNS, HTTP, TLS
Enough to trace "why will it not connect?"
Scripting
Bash or Python
Write scripts that automate repetitive work
Git
branch, merge, PR, tags
Required
One application
You can build a simple web app yourself
You need something to deploy in order to practice
YAML
Indentation, lists, maps
Writing CI configuration and Kubernetes manifests
Step-by-step roadmap: from Linux to operating Kubernetes
Step-by-step roadmap: from Linux to operating Kubernetes is a point where the options diverge. Using the table to compare what each approach is for and how it differs in operation makes later decisions much easier.
Step
Goal
What to learn
Guides
Done check
0. Linux
Working with servers
Shell, files and permissions, processes, systemd, SSH, network commands
12-week study plan: 1-2 hours a day is a point where the options diverge. Using the table to compare what each approach is for and how it differs in operation makes later decisions much easier.
The heart of the practice is to keep evolving one and the same application. After 12 weeks you have a complete pipeline in your portfolio that runs from "code push β automatic deployment β monitoring".
Week
Topic
Deliverable
Weeks 1-2
Linux commands, permissions, processes, networking
A server bootstrap script
Week 3
Docker images and the Dockerfile
A slim image of your app (multi-stage)
Week 4
Docker Compose
A local app + DB + cache environment
Weeks 5-6
CI/CD pipeline
GitHub Actions: test β build β push to the registry
Weeks 7-8
Basic Kubernetes resources
Deploy the app to a kind cluster + Ingress
Week 9
Helm + GitOps
A Helm chart + Argo CD auto-sync
Week 10
Prometheus + Grafana
A service dashboard + alert rules
Week 11
Cloud deployment
The same setup reproduced on managed Kubernetes
Week 12
Load testing + retrospective
A k6 load test report + an architecture document
Common mistakes of DevOps beginners
Common mistakes of DevOps beginners is a point where the options diverge. Using the table to compare what each approach is for and how it differs in operation makes later decisions much easier.
The golden rule of DevOps is "automate any task you repeat twice". Work done by hand leaves no record, and the next person cannot reproduce it.
Mistake
Problem
Do this instead
Starting with Kubernetes without Linux
Cannot trace the cause of an incident at all
Keep the order Linux β Docker β K8s
Deploying with the latest tag
No idea which version is running, cannot roll back
Use commit SHA or version tags
Putting passwords in images or the repository
Risk of leaks
A secret management tool, CI secrets
No resource requests or limits
One container hogs the resources of the whole node
Set requests/limits
No health checks
Traffic is sent to dead containers
Configure liveness/readiness probes
Operating without monitoring
Finding out about incidents from user reports
Set up dashboards and alerts together with the deployment
Changing server configuration by hand
An environment that cannot be reproduced
Put every change in code with IaC
Key terms
Key terms is a point where the options diverge. Using the table to compare what each approach is for and how it differs in operation makes later decisions much easier.
Term
One-line explanation
CI (Continuous Integration)
Building and testing automatically on every code change
CD (Continuous Delivery/Deployment)
Automatically deploying verified changes all the way to production
IaC
Defining and managing infrastructure such as servers and networks as code
GitOps
An operating model where the Git repository is the single source of truth for infrastructure and deployment state
Image registry
A repository that stores and distributes container images (Docker Hub, ECR, and so on)
Blue-green / canary deployment
Zero-downtime deployment that runs the new version in parallel or applies it to a slice of traffic first
SLI / SLO
A service quality indicator / the target for that indicator (e.g. 99.9% availability)
Autoscaling
Automatically adjusting the number of instances or Pods to the load
Postmortem
A blameless retrospective document that records the causes and improvements after an incident
Observability
The ability to understand the internal state of a system from metrics, logs, and traces
What should I study first to become a DevOps engineer?
Start with basic Linux commands and networking fundamentals, then widen out to Git, Docker, CI/CD pipelines, Kubernetes, and monitoring. Rather than learning each tool in isolation, building the whole flow yourself for one small application, build, deploy, and observe, is the most effective approach.
What is the difference between DevOps and SRE?
DevOps is a culture and set of practices in which development and operations share responsibility and ship quickly, and SRE is a concrete role and method that implements it with software engineering. What marks SRE is managing the balance between stability and release speed with numbers: service level objectives (SLOs) and error budgets.
Do developers need to know DevOps too?
You should know how the code you write is built, deployed, and observed. Understanding the constraints of running in a container, how configuration and secrets are injected, and how to emit logs and metrics lets you find the cause of an incident much faster.