How a Front‑End Publishing Platform Boosted Release Quality and Speed

Introduction

In the early stage KuJiaLe launched its front‑end tool platform (referred to as the "pub platform"). The platform was still in its initial construction phase and faced challenges such as poor release quality, low work efficiency, and a lack of emergency support tools, largely caused by a complex and cumbersome tool architecture.

When applying a gray‑release strategy, the team had to carefully consider plugin dependency links and the impact of generic business layer refactoring, as well as the tight coupling between front‑end and back‑end and the gray‑release time‑window requirements.

These complexities made the release process fragile: inconsistent integration node management reduced environment stability and compressed testing time, while frequent conflicts in gray‑release rules caused failures, leaving the team under constant pressure during each release.

Background

To improve overall release quality, the company started a tool release quality optimization project in 2022. After two phases of improvement, a unified application lifecycle management platform was built, offering rich release capabilities and a comprehensive quality‑gate system.

The core publishing platform, pub, consists of three modules:

Application Management : manages front‑end business applications and permissions.

Configuration Management : isolates configuration data from applications and supports online modifications.

Release Management : centralizes dependency orchestration, gray‑release rules, and reporting.

These modules enable standard releases, single‑branch releases, and gray releases.

Problem Analysis

KuJiaLe’s iteration process includes requirement, development, and testing stages. Early on, each stage relied on manual statistics and offline meetings, which became costly as the team grew and applications became more complex.

Three main problems were identified:

Chaotic application environments leading to high integration testing costs.

Single release strategy making risk control difficult.

Uncontrollable deployment quality and unstable test environments.

A standardized release chain and a flexible supporting platform are needed to ensure timely, high‑quality delivery.

Optimization Directions

3.1 Establish Admission and Exit Standards

The pub platform introduced a standard flow model (sit → beta → prod) with consistent standards and operations for each environment (development, testing, production).

Pre‑Release : cloud‑map permission rules, base library change standards, rotation‑log standards, whitelist management, new‑app onboarding, front‑end semantic naming.

During Release : cloud‑map release standards, prepare‑environment management, release rotation standards.

Emergency Operations : abnormal release standards, gray‑re‑publish standards.

Detailed operation documents (e.g., iteration release guide, emergency operation guide) were added to reduce learning cost.

3.2 Platform Usability Improvements

The pub platform now supports unified control of environments, resources, and configurations throughout the development cycle, offering visual workflow operations and five new release capabilities that solved most release difficulties of the past two years.

Key release scenarios and strategies include:

New feature gradual verification – gray release – risks: uneven user experience, potential data inconsistency.

Validate effects across versions – single‑branch release – risks: complex gray‑rule design, high implementation complexity.

Fast rollback to stable environment – latest backup release – risks: high resource consumption, configuration management complexity.

Multi‑dimensional phased plan – bucketed traffic – risks: deployment failures affect a subset of users, delayed feedback.

Release outside window – emergency release – risk: broad impact on all users.

3.3 Quality‑Gate Construction

While unified environments and diverse release strategies enable rapid delivery, some version instability remains. Automated quality gates were added to the deployment pipeline, improving environment stability and team collaboration. Frequent, reliable small‑scale changes allow the development team to respond quickly to demand changes.

Current quality gates include: prepare P0case gate, gray‑flow quality red‑line gate, pre‑dependency release gate, network‑closure change prohibition gate, etc.

Results and Outlook

4.1 Optimization Outcomes

Before optimization : gray‑release zero‑delay rate ~85%, on‑time rate only 73%; 45 version updates per year with 84 gray‑re‑publish occurrences.

After optimization : about 150+ applications and 30+ agile teams deliver product requirements weekly; 100% on‑time rate for the first half of 2024.

The pub platform now integrates with testing architecture, DevOps, monitoring, and enterprise workflow systems, forming a full‑ecosystem release solution. It also supports both domestic (KuJiaLe) and overseas (Coohom) product versions after a global‑process revamp in August 2024.

4.2 Future Outlook

Beyond unified environment management and diverse release strategies, the platform has built a flexible orchestration engine integrated with internal systems such as Moon, Kaptain, and iTest, and enforces strict permission controls.

Remaining challenges include high frequency of abnormal releases and long release cycles that rely heavily on manual operations, leading to errors and delays.

Future plans focus on deeper automation and intelligence: full‑pipeline automation from code commit to deployment, and leveraging machine learning and data analysis to mine historical release data for smarter decision‑making, aiming for a more efficient, stable, and flexible release system.