aliases: 
tags:
  - Type/Tech
  - area/tech
  - view/node/moc
publish: true
permalink: devops
version: 1
dateCreated: 2021-10-20, 09:49
dateModified: 2024-08-08, 15:44
from:
  - "[[Tech]]"
related:
  - "[[Programming]]"
  - "[[Tech]]"
  - "[[Site Reliability Engineering]]"
  - "[[Stages of DevOps]]"
  - "[[+CI CD]]"
  - "[[DevOps Tools]]"
  - "[[Software Measurement]]"
  - "[[Agile Software Development]]"
  - "[[Project & Product Management, Ownership, Leadership, & Agile]]"
  - "[[Testing & QA]]"
  - "[[Cloud]]"
  - "[[Security]]"
contra: []
to: []

DevOps

	DevOps is a methodology in the software development and IT industry. Used as a set of practices and tools, DevOps integrates and automates the work of software development (Dev) and IT operations (Ops) as a means for improving and shortening the systems development life cycle.

The Gist

Portmanteau of developer and operations
The practice of linking and integrating software development and Operations/IT
Essentially an outgrowth of Agile Software Development and Lean

Concepts & Principles

DevOps represents a change in IT culture, focusing on rapid IT service delivery through the adoption of agile, lean practices in the context of a system-oriented approach. DevOps emphasizes people (and culture), and seeks to improve collaboration between operations and development teams. DevOps implementations utilize technology— especially automation tools that can leverage an increasingly programmable and dynamic infrastructure from a life cycle perspective.

3 Different Definitions of DevOps

The concepts and philosophy, the goals, the unification of ops and dev, attaining faster delivery of higher quality code
The role of DevOps - DevOps Engineer
- Understanding DevOps Roles and Responsibilities
The tools and processes of DevOps - pipelines, monitoring, etc.

People Over Process Over Tools

Don't decide the tool and work backwards to what process and people are needed.

Breakdown Traditional Silos and Competing Incentives between dev and ops

dev (deliver features fast) and ops (stability - manage risk by fewer changes to prod)
Kind of emerged as the unification of two otherwise competing demands - slow and stable IT operations vs quick and adaptive development.
Don't just throw it over the wall - we are all responsible for the whole process - we share the goals of both speed of delivery of features plus stable production uptime.
Avoid the black box nature of silos and non-collaborating teams that tends toward finger pointing and mistrust

Embedded, Cross-functional Teams

E.g., embedding ops people on a dev team
Greater collaboration among teams

The Three Ways from Gene Kim

Systems Thinking, Feedback Loops, & Continuous Experimentation

CALMS DevOps Framework

Chaos Engineering & Antifragility

Andon Cords

Kaizen From Lean which is from Agile Software Development

Visible Ops

non-fragile artifacts, repeatable builds, etc.

More of a culture than a particular role/position, methodology or tool.

Blameless Postmortem

Knowledge Sharing

Consensus-based DevOps

[How to achieve DevOps consensus: The what and how of DevOps · Raygun Blog](https://raygun.com/blog/what-is-devops/

Best Practices

Keep builds and deploys fast

Each developer is responsible for their check-in through deployment to prod

Public Static Pages and build statuses

Like those build passing green light things on open source package pages
Public status pages and public build success/verification/tests
Have a slack channel that sends the deployment status at each stage.
be public and honest about outages, and be open and clear with what went wrong with your customers

Artifacts should be immutable.

e.g., remove write access to the package in artifactory
People need to have no doubts that the artifact is identical
Deployments should go to a copy of production (staging)
Deployments should be idempotent.
Stop deploy if previous step fails.

Shift Left

Test fixtures are artifacts so they should be in version control and a repo

Broken builds/own your build

Don't check in broken builds, own your build when it breaks

Automate deployments and builds

Shouldn't depend on any person or machine to do a build or deployment

Don't ignore broken tests

Always easy, quick way for a dev to rollback any given deployment to handle issues on prod that often arise from a deployment

Simple, trunk-based branching strategy (master/main) and minimize branching

Trunk-based Development v
Git

Only build artifacts once.

Agile style small, iterative changes

Deployments should always self-validate. At least do a smoke test.

Golden Image or foil ball

The idea that the initial base/template image should be as minimal as possible and let your configuration management do as much configuring as possible and while using infrastructure as code.