Why Process Management Is the Secret to DevOps Success in Modern Enterprises

Preface

According to the law of natural development, any thing that has a continuous drive for evolution will inevitably pass through stages of automation and intelligence.

In the Internet industry, to stand out one must manage the fundamentals of process management, which can greatly reduce internal costs and produce high‑quality products externally. This article focuses on the methodology and toolchain of process management within an Internet technology system.

Philosophical Background

Human beings, as products of nature, follow this evolutionary law. Unlike other species, humans can create tools that liberate productivity. Today we can travel to space, cross oceans, and seemingly do anything, yet scientists still dream of creating artificial intelligence identical to humans.

Biological intelligence is a product of time, not technology, so true human‑level AI is impossible without controlling time. The driving forces behind humanity’s progress are survival instinct and self‑awareness, the latter manifesting as curiosity.

Technical Implementation vs Process Management

Technical Implementation is the theory that achieves a goal; Process Management is the production and operation process guided by that theory.

Analogy:

Cat : the cat’s paw represents technical implementation; silently approaching the mouse represents process management.

Human : designing a mouse trap is technical implementation; selecting materials, bait, and placement is process management.

Both the cat and the human need both aspects to catch the mouse, but only humans embed methodology into tools, which is why humans differ from animals.

In the Internet industry, mastering process management (the “root”) and technical implementation (the “branch”) is crucial for cost savings and high‑quality product delivery.

Two Directions of Process Management

Execution System : the production process that achieves the final goal.

Observation System : continuous monitoring after the product goes live.

Example: an aircraft’s manufacturing is the execution system; its flight parameters monitoring is the observation system. Online products are like aircraft flying in the sky – they need both advanced production techniques and advanced monitoring.

Platform Cases

Execution System: DEV&AIOps Part 1 – Overmind Implementation (this article)

Observation System: DEV&AIOps Part 2 – Alarm Platform Construction (to be continued)

Part 1: Theory

First, explain the theoretical basis for why we do something, not the concrete practice.

Both process management and technical implementation aim to serve the end user. From low to high levels:

Process Management Levels

Low: low efficiency, high cost.

Medium: high efficiency, low cost.

High: collaborative output, enabling products such as DingTalk and Enterprise WeChat.

Technical Implementation Levels

Low: poor systems due to weak process and technology.

Medium: good systems from efficient process and solid technology.

High: external technology output forming a tech market (e.g., Alibaba Cloud, Tencent Cloud).

Advanced process management is collaborative output, a modern upgrade of traditional process management.

Enterprise drive roadmap:

From high to low:

Technical‑Operations drive: integrated technical‑operations.

Technical‑Product drive: product managers with technical background.

Business drive: technology follows business.

Client drive: passive response to client needs.

Leadership drive: planned economy.

The source of enterprise progress is twofold: biological survival instinct and curiosity‑driven self‑awareness. In business this appears as profit motive; the evolution from profit‑driven to curiosity‑driven models creates interest‑driven enterprises.

Interest in Practice

Interest manifests in two processes: creating technical equipment to achieve goals, and using refined equipment to capture targets (e.g., fishing).

When interest combines with survival, the resulting gains fund better equipment, creating a virtuous cycle.

Technology‑Driven Enterprise Model

Technology is the primary productive force. A technology‑driven enterprise treats technology as the engine, aligning internal efficiency with external ToB collaboration.

Case Study: Delivery Logistics Outsourcing

Business‑Driven : manager sets cost‑reduction KPI; solution relies on negotiation, yielding high cost and instability.

Technology‑Driven : manager asks for a sophisticated logistics system; technical team designs a system with clear benefits.

Benefits of the tech‑driven solution (six points): single panel, high‑efficiency interfaces, pricing power, competitive bidding, elimination of manual waste, optimal cost for all parties.

Comparison of Major Platforms

Tencent – Zhiyun One‑Stop Platform

Integrates Enterprise WeChat, TAPD, etc., achieving full‑chain collaboration across development, testing, operations, and users.

Alibaba – CloudEff (Aone) One‑Stop Platform

Integrates DingTalk, Teambition, and other tools for project and development collaboration.

Baidu – Single DevOps Suite

NetEase – Current Status

No ToB ambition; internal collaboration tool.

No self‑built project‑level platform (uses Jira).

Overmind is a mature internal product improving production efficiency.

Monitoring and alarm are relatively simple compared with BAT’s AIOps.

Methodology

DevOps feedback loop consists of four principles and four measurement criteria.

Four Principles

Rapid problem discovery.

Problem analysis and resolution.

Source‑level quality assurance.

Customer empathy.

Four Metrics

Engineering efficiency: time from requirement to launch.

Stability: high efficiency must not sacrifice stability.

Non‑R&D work ratio: higher ratio reduces success probability.

Ops‑to‑service ratio: e.g., Google’s 1 ops per 2000 machines.

2018 international data shows the potential of a mature DevOps system to cut waste and improve productivity.

Implementation Steps

Tool Unification : same tools across the organization.

Process Unification : consistent workflows to avoid hidden costs.

Architecture Unification : resolve language, protocol, and framework differences.

When all three are unified, every member works toward the same business goal with a shared technical language.

Overmind Implementation Details

Adaptations:

Integrated GitLab‑CI.

Supported trunk‑based and branch‑based release models.

Online branch merging.

Quality data platform.

Improvements:

CI: GitLab‑CI automates development‑ops feedback.

CD: Opera replaces NDP for artifact‑based automated deployment.

Configuration Management: unified via Apploy (still evolving).

Monitoring & Alarm: fragmented but converging toward unified AIOps.

These feedback loops belong to the observation system, to be detailed in the sister article.

DevOps Principles Re‑emphasized

Fast problem discovery via Jira, CI, CD, post‑deployment checks, and observation system.

Automated assignment and workflow interruption for problem analysis.

Source‑level quality guarantees through Jira (requirements), CI (code), layered testing (product), CD (deployment), and monitoring/alarms (runtime).

Customer empathy: user‑oriented development and integrated Dev‑Ops.

Finally, the article concludes that true AI comparable to biological intelligence is unattainable, that enterprises should shift from profit‑driven to interest‑driven models, and that technology‑driven, DevOps‑enabled organizations are the future.

Boss: “Why does my department seem useless?” Ops Manager: “Have you crossed a bridge?” Boss: “Yes.” Ops Manager: “Is there a railing?” Boss: “Yes.” Ops Manager: “Do you hold the railing while crossing?” Boss: “No.” Ops Manager: “Then the railing is still useful, right?” Boss: “Of course, without it I’d fall.” Ops Manager: “Exactly, ops is the railing.”

Author Bio

Liang Shu, former engineer at Shanda and Tencent, now a technical expert at NetEase Yanxuan, specializing in backend architecture, e‑commerce supply chain, performance optimization, and efficiency systems.

本文由作者授权严选技术团队发布