Opscraft
DevOps from zero · Lesson 1 of 6 · 5 min read

What DevOps is and what a DevOps engineer does

Start here if you are new. This topic assumes no background at all. It explains what DevOps is, what actually happens when someone opens a website, what servers, virtual machines, containers and "the cloud" are, how to use a command line, and how code travels from a developer's laptop to real users. Every later topic builds on these ideas, and the last lesson shows the path from beginner to Senior DevOps engineer.

The problem DevOps solves

Most software companies have two kinds of work:

  • Development (Dev) — writing new features. Developers are rewarded for shipping changes.
  • Operations (Ops) — keeping the running system up. Operations teams are rewarded for stability, so every change looks like a risk.

When these are separate teams, code gets "thrown over the wall": developers finish a feature, hand it to operations weeks later, and nobody fully understands the other side. Releases become rare, big and scary. When something breaks at 2 a.m., the people woken up are not the people who wrote the code.

Enlarge diagramSEPARATE TEAMSONE DEVOPS TEAMDeveloperswant to shipOperationswant stabilityYou build it, you run itcode, pipelines, servers and on-call togethercode is handed over the wallBig, rare, risky releasesFeedback takes weeksBlame when things breakautomation connects every stepSmall, frequent, boring releasesFeedback in minutesFix the system, not the person
Figure 1. Before DevOps, code was "thrown over the wall" between two teams with opposite goals. DevOps puts building and running a service in the same hands.
Swipe sideways to see all of it, or tap to open it full screen.

What DevOps means

DevOps is a way of working, not a single tool or job title. The same team owns a service from writing the code to running it in production, and it automates every repeatable step in between. It rests on a few ideas:

  • Small, frequent changes. Shipping ten small changes is safer than one huge one, because each is easy to test, understand and undo.
  • Automation. Builds, tests, deployments and even servers are created by code, so they happen the same way every time.
  • Fast feedback. Every change is tested within minutes, and monitoring tells you quickly if users are affected.
  • Shared ownership and blameless learning. When something fails, the team fixes the system that allowed it, instead of blaming a person.

These steps repeat forever, which is why DevOps is often drawn as a loop:

Enlarge diagramPlandecide what to buildCodewrite it, get itreviewedBuildcompile and packageTestautomated checksReleaseapprove a versionDeployroll it out safelyOperaterun, scale, patchMonitorwatch users'experience12345678learn and improve
Figure 2. The DevOps loop. Development steps (blue) and operations steps (green) repeat continuously; what monitoring shows feeds the next plan.
Swipe sideways to see all of it, or tap to open it full screen.

What a DevOps engineer does day to day

Job titles vary, but the work is similar. A typical week might include:

  • Building and fixing CI/CD pipelines — the automation that builds, tests and deploys code (GitHub Actions, Jenkins).
  • Creating cloud infrastructure with infrastructure as code (Terraform, CloudFormation) instead of clicking in a web console.
  • Running containers and Kubernetes clusters that host the company's services.
  • Setting up monitoring and alerts, and joining an on-call rotation to respond when something breaks.
  • Managing security: who can access what, where passwords and keys are stored, patching.
  • Helping developers ship faster: templates, documentation, and fixing "it works on my machine" problems.
  • Keeping an eye on cloud cost.
Title Focus
DevOps engineer Pipelines, infrastructure as code, cloud, containers — the broad middle ground
Site Reliability Engineer (SRE) Reliability measured with numbers (SLOs), incident response, removing repetitive manual work
Platform engineer Building an internal "platform" (templates, self-service tools, Kubernetes) that other teams use
Cloud engineer Designing and running cloud accounts, networks and services

In job descriptions these overlap heavily. This site covers the skills common to all four.

How teams measure DevOps

The DORA research programme (now part of Google Cloud) found four numbers that separate high-performing teams from the rest:

  1. Deployment frequency — how often you release to production.
  2. Lead time for changes — how long from a commit to that code running in production.
  3. Change failure rate — what share of releases cause a problem.
  4. Time to restore service — how quickly you recover when a release breaks something (newer DORA reports call it failed deployment recovery time).

Good teams are fast and stable at the same time. Speed and safety are not a trade-off when changes are small and automated.