How Tencent’s Blue Whale Transforms Operations: From Automation to Data‑Driven Service

Introduction

Recently the concept of "Operations 2.0" was proposed by Mr. Xiao of Touch Technology, echoing the earlier "Operations Transformation" project at Tencent Games. The project, embodied by the Blue Whale system, has reshaped application operations (ARE) at Tencent Games and now supports roughly half of the domestic gaming market.

Blue Whale is the codename for Tencent Games' application operations technology ecosystem, consisting of six gradually productized platforms and numerous DevOps and operational planning personnel. It aims to equip operations teams with higher‑dimensional services, such as self‑service tools, data‑driven decision support, and direct user‑experience improvements.

This background and overview were prepared for a live sharing session on July 16, intended to help peers find relevance in their own stages of operations transformation.

1. Background: Operations Transformation

Ten years ago, operations focused on servers, networks, OS, DB, releases, changes, monitoring, fault handling, and environment data extraction—mostly passive, demand‑driven tasks with heavy repetition.

Five years ago a pilot team attempted to shift from "operational service output" to "solution service output". Three years later, after evaluating the pilot, over a hundred operations teams (now 200+) across Tencent Games embarked on a difficult transformation, with Blue Whale as the implementation platform.

Reason 1 – Business Red‑Ocean

Intense competition demands fine‑grained operations. Product, planning, and development teams focus on user‑centric design and rapid delivery, while operations strive for near‑continuous availability and provide tools and data to improve product‑operation efficiency.

Reason 2 – Shrinking Space for Traditional Operations

Emerging technologies reduce the need for classic operations tasks; development teams increasingly handle fine‑grained operational work themselves, threatening traditional operations roles.

Reason 3 – Operational Fatigue

The massive growth of Tencent Games made basic tasks like releases and fault handling unsustainable without transformation.

Thus the long‑term goal of operations transformation is to automate basic services (releases, monitoring, data extraction) to be unattended, provide solution tools for self‑service or outsourcing, and allocate effort to user‑experience optimization and decision‑support services.

2. Design Philosophy

Tencent Games faces four inherent challenges: diverse business types (client, web, mobile, platform), a wide variety of technologies from many developers, no uniform operational workflow across games, and a massive scale of servers (hundreds of thousands).

Consequently, Blue Whale must be non‑intrusive, technology‑agnostic, and independent of any unified operational process, supporting up to ten‑thousand concurrent operation units.

The core insight is that all operational steps can be performed via Linux commands, which can be abstracted into atomic actions. Blue Whale automates these atoms and connects them through a task engine to form linear or branching workflows, enabling full automation without relying on business specifics.

The two key tasks are:

Automating atoms: wrapping command‑line steps into scripts on the Blue Whale Job Platform or integrating UI‑driven steps via the ESB platform.

Aggregating atoms into tools: providing a PaaS module where developers write logic once and deploy to containers, a unified code framework with login, permissions, and tagging, and a front‑end component library to reduce UI development effort. Training enables new graduates to build APP tools on the integration platform within four weeks.

3. Phased Evolution

Stage 1 – Automating Basic Operations

Automate repetitive, environment‑triggered tasks (scaling, opening servers, merging, alarm handling) to achieve unattended operation, allowing teams to sleep at night.

Stage 2 – Enabling Product‑Driven Automation

Encapsulate product‑initiated actions (releases, configuration changes, data extraction) into self‑service APP tools, letting product teams operate without waiting for operations.

Stage 3 – Data‑Driven Operations

Integrate real‑time big‑data capabilities (Kafka, Storm) into the Blue Whale Data Platform, allowing operations to write YAML‑described logic for data collection and processing, and display results via APPs, thus providing decision‑support analytics.

4. Platform Architecture

Blue Whale comprises six platforms:

Integration Platform (PaaS, ESB, development framework, web samples) – the tool‑building hub.

Mobile Platform – mobile entry point.

Job Platform – file transfer and script execution.

Configuration Platform – hierarchical storage of business attributes.

Control Platform – unified agent for servers, containers, and big‑data pipelines.

Data Platform – Kafka‑Storm based real‑time analytics for decision‑support tools.

The Data Platform offers an online IDE with YAML, connects to diverse operational data sources, and provides a data dictionary for custom analytics.

5. Services: PaaS vs SaaS

All previously described services are PaaS: developers create APP tools, write scripts, or define data pipelines themselves. PaaS offers unlimited flexibility but can lead to duplicated effort. To address common scenarios, Blue Whale also offers SaaS solutions such as the "Standard Operations" APP for release‑change workflows and a generic "Fault Self‑Healing" service that packages alarm handling and recovery components, with a Python‑based editor for complex fault trees.

6. Conclusion

Operations has long been undervalued, yet Blue Whale demonstrates how empowering operations with automation, self‑service tools, and data analytics can create high‑value, service‑oriented capabilities that complement product, planning, and development teams. Through continuous refinement of its six platforms, Blue Whale aims to support Tencent’s internal and partner businesses, and to foster a collaborative ecosystem for application operations.