Reflections on DevOps Organizational Transformation: Lessons from Development‑Operations Integration, Product Teams, and IT Ops Decentralization

“In the past two years, both the group and departments have been vigorously undergoing a DevOps organizational transformation to improve delivery efficiency and respond faster to business needs. However, new problems have emerged, and continuous improvement is an endless journey.”

Lesson: Others' successes cannot be copied, but you can avoid the pits they fell into.

01 – Development/Operations Integration

Previously, development and operations teams were separate, with a consolidated ops team handling frontline incidents and bearing the blame. Developers focused on project delivery, while ops lacked involvement in development, limiting knowledge sharing. Following the DevOps principle of "who develops, who operates," the ops team was disbanded and its members were embedded into development squads per business line, aiming to give end‑to‑end responsibility and increase satisfaction.

This model introduced two new issues:

Because personnel rotate, the original developer may leave, making it hard to hold anyone accountable for later incidents.

For global business spanning time zones, the dispersed ops model hampers shift hand‑offs and forces teams to work outside regular hours, affecting next‑day delivery.

Lesson: When supporting global services, consider whether merging dev and ops is truly the right approach.

02 – Product Team

Traditional project‑based teams do not address ongoing product evolution, technical debt repayment, or continuous DevOps improvement. Therefore, teams were reorganized by system boundaries, each small team owning end‑to‑end delivery—including development and operations—and managing its own backlog.

Goals of this restructuring include:

People: Everyone can work on different projects, avoiding long‑term siloed maintenance roles and boosting morale.

DevOps: Shifting from a project view to a product view initiates continuous product improvement.

Delivery: More effective staffing for varied project needs.

Agile: Each squad manages its own backlog and iteration plan.

Operations: Everyone can contribute to production, fulfilling the "who develops, who maintains" principle.

Responsibility: Clear end‑to‑end ownership reduces hand‑offs.

Technology: Facilitates architecture review and refactoring.

However, conflicts arose between project managers and product owners. Project managers, still using waterfall planning, demanded detailed schedules, but the squads no longer reported to them, leading to disputes over who should create plans. Additionally, the lack of a business‑side Product Owner (PO) caused misalignment, as the new team structure is based on IT systems rather than business products.

Lesson: To implement product/feature teams, first identify a reasonable business‑product split that is small and independent, then align IT teams with it according to Conway's Law.

03 – Decentralization of the IT Operations Department

The former centralized ops department controlled all infrastructure (servers, middleware, network) and held critical permissions, becoming a bottleneck as business demands grew. Although efforts were made toward private‑cloud and self‑service, progress stalled.

Decentralizing ops by redistributing staff to business line teams introduced new dilemmas, such as determining the appropriate size for a full‑time DBA. Small teams may under‑utilize a DBA, while sharing a DBA across teams recreates the original bottleneck, especially when time‑zone differences are involved.

Lesson: Balancing responsiveness and resource efficiency is a persistent trade‑off; cloudification and self‑service of IT ops may be the key focus of reform.

All the problems described are the outcomes of previous solutions; continuous improvement remains essential.

About the Author

Early agile practitioner

Started with Extreme Programming

Experienced in XP, Scrum, Kanban, TDD, CI, BDD, and DevOps toolchains

Author of "Cheetah Action: The Agile Transformation Journey in the Smoke of Battle"

---

DevOps+LIVE Live Broadcast Announcement

Topic: "Linux Containers on Windows and Kubernetes in Azure Private Cloud" – Live on 21 Nov 2018, 20:00.

Speakers:

Peng Aihua – Microsoft MVP for Azure Stack, expert in hybrid cloud, author of many Windows books, will discuss Linux containers on Windows and Kubernetes inside Azure Stack.

Alan Liu – Microsoft MVP, chief architect at Macro Cloud, focuses on Azure Kubernetes implementations, will share hands‑on experience from the field.

To register, follow the DevOps public account and send the keyword 加入社区 to be added to the live‑share group.