The Historical Conflict Between Development and Operations and the Evolution of DevOps

In the article "#DevOps的前世今生#2. Dev和Ops矛盾缘何而来", the author reviews the historical tension between development (Dev) focused on rapid, agile delivery and operations (Ops) centered on stability and performance, highlighting how this conflict was first articulated by John Allspaw and Paul Hammond in their "10+ Deploys Per Day: Dev and Ops Cooperation at Flickr" presentation.

The speakers challenged the traditional view that Dev adds features while Ops ensures stability, proposing instead that Ops should aim to "enable the business"—a goal aligned with Dev. They argued that activating business leads to more changes, which in turn increase failure risk, creating a trade‑off between stability and rapid change.

To resolve this, Flickr adopted a cooperative approach that combined tooling and culture: automated infrastructure, shared version control, feature flags, one‑step build and deploy, frequent small changes, shared metrics, and real‑time communication tools (IRC/IM bots). The cultural shift emphasized mutual respect, trust, and a blame‑free attitude toward failure.

The article then outlines four feedback loops that illustrate DevOps practice: (1) Development‑Testing loop driven by non‑functional requirements such as scalability; (2) Development‑Operations loop where ops adopt software development techniques (code repositories, CI, configuration management, automated testing, monitoring integration); (3) Business‑Operations loop leveraging cloud and agile infrastructure, business‑level non‑functional concerns, and ops participation in upstream processes; (4) Business‑User loop where ops act as the first responders to user issues, feeding back to the business.

Key technical upgrades and process management practices identified include infrastructure automation, shared version control, feature flags, one‑step build/deploy, small frequent changes, shared metrics, and integration of chat/IM tools. These technical enablers, together with a collaborative culture, aim to embed quality into the software delivery pipeline and accelerate the value stream.

DevOps is described as adding a feedback loop between operations and the value stream, thereby improving quality and speeding up end‑to‑end delivery. The article notes challenges such as the vague definition of DevOps and resistance from traditional ops, but emphasizes ongoing refinement of the cultural and technical aspects.

In conclusion, DevOps is presented as a technical and managerial transformation that embeds quality to accelerate software value flow, while acknowledging that its broad scope and lack of a unified methodology can cause confusion and uneven adoption across organizations.