How Alibaba’s ‘Ali Inside‑Out’ Uses Cloud‑Native Architecture for Collaboration

Overview of Ali Inside‑Out

Ali Inside‑Out is the enterprise operation and collaboration platform used by Alibaba employees. Launched in 2013 as a portal and social entry, it has grown into a platform‑based service that integrates numerous Alibaba applications and provides unified services for the ecosystem.

Cultural and Collaborative Features

Ali Inside‑Out embeds Alibaba’s cultural elements, such as the “Ali Flavor” section where executives and employees share ideas, discuss organizational issues, and post technical or product insights. The platform also offers a suite of collaboration tools—Q&A, task tracking, weekly reports, documents, and team workspaces—accessible through a one‑stop search that aggregates work content and visualizes yearly achievements.

Technical Architecture

Technology Stack

The platform builds on Alibaba’s extensive tech stack, reusing components like Spring‑Boot, Spring‑MVC, Hystrix, and Alibaba‑specific middleware (AliOS, AliJVM, AliTomcat). It adopts a “three‑horizontal, two‑vertical” model: the top horizontal is a unified HTTPS entry layer; the middle horizontal is the application layer with Spring‑Boot and various plugins; the bottom horizontal reuses group‑level middleware such as TDDL, Tair, MetaQ, OTS, and Redis. The two verticals cover DevOps operations and monitoring (system, application, front‑end, node) using Alibaba’s full‑link monitoring suite EagleEye and log collection systems.

Search Framework

Ali Inside‑Out’s search architecture consists of three layers. The data layer connects content sources via offline ODPS dumps, API‑driven interfaces, or direct database dumps. The middle layer uses Alibaba’s self‑developed ISEARCH engine and the NEO4J graph database to index documents, sites, and organizational structures. The top layer performs query intent recognition and supports various application scenarios, leveraging algorithms from the PAI platform for tokenization, spelling correction, and suggestion features. Real‑name login enables personalized “thousand‑person‑one‑face” results.

Heat‑Based Feed Stream

To improve information focus, Ali Inside‑Out replaces traditional forum ordering with a heat‑driven feed. Posts are classified by an algorithm, stored in a database, and their interaction logs are fed into a computing platform that calculates popularity scores. The scores are cached in Redis and used to rank content dynamically based on user browsing habits.

Global Deployment

With nearly 60,000 employees worldwide, Ali Inside‑Out leverages Alibaba’s global ADNS service and cloud‑based IaaS infrastructure to provide low‑latency access for overseas staff. Data consistency is ensured through Alibaba Cloud Data Transmission Service (DTS) for backup and synchronization, delivering a “silky‑smooth” experience regardless of geographic location.

Future Outlook

Currently restricted to Alibaba internal employees for security reasons, Ali Inside‑Out is gradually optimizing its product to expose selected functionalities to external users via Alibaba Cloud or DingTalk, aiming to share its collaborative advantages beyond the company.