How a $100B Financial Giant Mastered DevOps: Real‑World Transformation Lessons

Foreword

The talk is based on a DevOpsDays Beijing presentation that shares practical experience of applying DevOps in a traditional financial organization.

Background

The organization is a domestic financial asset group with $100 billion in assets, extensive insurance and banking businesses, and a fragmented IT landscape caused by years of acquisitions.

Challenges Faced

In 2009 the company struggled with intense market competition, outdated online sales systems, duplicated legacy applications, talent loss, and inefficient organizational structures.

Transformation Path

Since 2006 the company began an agile transformation that evolved through three major phases:

Organizational restructuring – consolidating business lines and unifying operations.

Legacy system optimization – reducing dozens of customer‑management and claims systems to single instances and standardizing digital channels.

Building continuous delivery – establishing a version‑controlled code repository that stores application code, tests, build scripts, and deployment scripts.

Product Team Delivery Pipeline

A continuous integration service triggers build scripts, creates a gated admission mechanism, and publishes artifacts to a central repository. Deployments are performed via one‑click buttons to multiple environments (int, A, B) with integrated monitoring, alerting, and user‑experience metrics displayed on dashboards.

Remaining Issues in Large‑Scale Continuous Delivery

Balancing stability and throughput – manual release processes and night‑time deployments cannot keep up with rapid change cycles.

Fostering a business‑value culture – developers often lack direct feedback on user satisfaction, leading to delayed value verification.

Controlling growing system complexity – siloed development and operations increase maintenance costs and hinder delivery.

Typical Example

A development team requests a set of machines for build, test, and production environments; automation scripts provision the resources, reducing manual OS, network, and middleware setup.

What Is DevOps?

DevOps is introduced as a cultural and technical approach that bridges development and operations, enabling faster, reliable delivery.

DevOps Implementation

The organization’s practice is divided into four stages:

Introducing dynamic infrastructure (cloud computing).

Restructuring the IT department and establishing self‑service mechanisms.

Adopting Infrastructure‑as‑Code (IaC).

Building a platform‑as‑a‑service ecosystem.

Step 1: Dynamic Infrastructure

Resources such as compute nodes, load balancers, networks, and IPs are provisioned on demand via APIs, SDKs, or scripts, effectively leveraging public cloud while maintaining data security through hybrid architectures.

Public Cloud Boosts Productivity

Migrating to public cloud cut disaster‑recovery costs by 85 %, eliminated the need for on‑premise data‑center maintenance, and enabled developers to provision virtual desktops in minutes, providing high‑availability and elastic storage for large‑scale logging workloads.

Public Cloud Migration Strategy

The migration follows a principle‑driven design: isolated workspaces for non‑prod and prod, integration with existing data centers, and a phased approach prioritizing low‑risk, high‑value applications.

Private Cloud Within Public Cloud

Integrate corporate SSO with cloud authentication.

Establish direct VPN connections between on‑premise and cloud resources.

Migrate low‑risk, high‑value services first, keeping critical data on‑premise.

Step 2: Organizational Restructuring & Self‑Service

A dedicated DevOps team was created to provide self‑service portals, allowing developers to modify firewall rules or request resources without manual ops intervention, while an approval workflow ensures security and accountability.

DevOps Team Role

The team bridges development and operations, fostering empathy, automating routine tasks, and freeing ops staff for proactive optimization.

Team Structure

The DevOps tribe holds regular knowledge‑sharing sessions and drives continuous improvement.

Team Responsibilities

Maintain the continuous delivery ecosystem.

Design and validate disaster‑recovery and deployment strategies.

Improve productivity by automating idle resource shutdown.

Empower product teams through training and tooling.

Step 3: Infrastructure‑as‑Code

IaC treats infrastructure, tools, and services as software, using DSLs to declaratively describe environments, storing all definitions, scripts, and application code in version control for reproducible builds.

DSL‑Based Environment Description

Developers write simple files that specify machine size, software packages, and network settings; the pipeline automatically creates the required environments.

Everything in Version Control

Application code, test suites, build scripts, database migration scripts, and deployment scripts are all versioned, enabling fully automated deployments without manual intervention.

Step 4: Platform‑as‑a‑Service

The organization builds a unified platform offering self‑service APIs for logging, monitoring, and resource provisioning, preferring commercial or open‑source solutions over custom development.

Platform Principles

Provide professional‑grade services, align with developer pain points, and continuously optimize performance (e.g., log search speed).

Summary – Core Elements of DevOps Adoption

Self‑service with review mechanisms.

Automation of all repeatable tasks.

Knowledge sharing to upskill all participants.

DevOps Is Not a Silver Bullet

Successful adoption requires focus on people, principles, timing, product mindset, and adaptable processes rather than expecting a one‑size‑fits‑all solution.