The Evolution of DevOps: From Agile Foundations to CALMS, Containerization, and Enterprise Best Practices

In 2008 at the Agile conference in Toronto, Patrick DeBois and Andrew Clay Shafer first proposed discussing "Agile infrastructure". The subsequent "Deploying 10 times a day" talk from Flickr sparked the first DevOps Days in Ghent, Belgium in 2009, where Patrick DeBois publicly coined the term "DevOps". He is now often called the "father of DevOps".

At the 2010 DevOps Days in Mountain View, Damon Edwards introduced the acronym "CAMS" (Culture, Automation, Measurement, Sharing). Later, Jez Humble added Lean principles, forming the extended "CALMS" model.

Culture – embracing change and fostering collaboration and communication.

Automation – removing manual interventions from the value stream.

Lean – applying lean principles to enable high‑frequency cycles.

Metrics – measuring each step and using data to improve the loop.

Sharing – openly sharing successes and failures to learn continuously.

The CALMS model aligns with DeBois’s view that “DevOps is a human problem”.

With the rise of Chinese internet giants such as Baidu, Alibaba, and Tencent, domestic DevOps practices have been widely shared at local conferences. Over the past two years, many Chinese enterprises have begun experimenting with DevOps, often focusing on automation, containerization, and PaaS platforms.

Docker container technology has dramatically lowered the barrier to continuous delivery, reshaping software supply chains. Container‑based micro‑service architectures have become increasingly popular, prompting traditional enterprises to pilot DevOps initiatives that range from infrastructure containerization to automated delivery pipelines. Most efforts remain at an early, exploratory stage.

The article classifies DevOps evolution into 1.0 and 2.0 phases and compares them with traditional approaches across several dimensions.

Comparison

Traditional

DevOps 1.0

DevOps 2.0

Software Architecture

Monolithic applications

Distributed systems

Micro‑service architecture

Development Cycle

Releases measured in months or years

Weekly releases

Hourly or minute‑level updates

Code & System Quality

Single‑language, tightly coupled components

Heterogeneous systems, language diversity, database quality critical

Polyglot services, high code quality per service, independent scaling and fault isolation

Deployment Difficulty & Time

Month‑long, manual, high risk

VM‑based updates via templates, reduced difficulty, heavy rollback effort

Container‑based releases supporting blue‑green, canary, easy rollback

Implementation Cost

High development, deployment, and operations cost

Agile improves efficiency; CI/CD and automation lower deployment and ops cost

Initial development investment high, but after mastering micro‑services costs drop and time to deploy is dramatically optimized

Three technologies dominate current Chinese DevOps practice:

Containers – eliminate environment dependency, speed up deployments, and lower cost; Docker runs on any Linux host with minimal footprint, but requires careful planning of resource allocation, networking, and image registry selection.

Continuous Deployment – standardized automated pipelines (e.g., Jenkins) enable repeatable, environment‑agnostic releases, supporting blue‑green and canary strategies.

Micro‑services – refactor monoliths into loosely coupled services, a natural complement to containers and continuous delivery.

These three aspects reinforce each other: containers provide the lightweight, immutable runtime that fuels continuous delivery and micro‑service architectures.

Best‑practice recommendations for enterprise DevOps include:

Agile Management – adopt agile methods for planning, design, and development; keep work items small, enforce continuous integration, and shorten release cycles to increase feedback.

Continuous Delivery – automate build, test, and deployment pipelines; embed automated testing to ensure repeatable, low‑risk releases.

ITSM Service Management – shift traditional ITSM toward lightweight, business‑continuity focus; involve operations early to address non‑functional requirements.

Lean Management – apply lean principles across the lifecycle, eliminating waste, enabling flow, and fostering continuous improvement.

Implementing DevOps in large, traditional enterprises is complex and requires both top‑down support and grassroots innovation, often borrowing lean manufacturing practices from production industries. Nevertheless, DevOps is poised to have a significant impact on the IT sector in the coming years.

(Source: Exin DevOps Whitepaper)