How Meituan Scaled Its Tech Architecture from LAMP to a Cloud‑Native O2O Platform

Introduction

Meituan’s technology committee chair Xia Huaxia shares the company’s technical journey, emphasizing that continuous technical effort and a pursuit of excellence are key to Meituan’s success as China’s largest local‑life service platform.

Technical Architecture Evolution

In the early years (2010‑2011) Meituan used a classic LAMP stack: Apache, PHP, MySQL, with basic operations and simple caching (Varnish, Memcached). As traffic grew, the architecture was incrementally optimized, adding Nginx, expanding caching layers, and introducing a mobile API layer in 2011.

When new business lines (hotels, movies, food delivery) were launched, Meituan extracted common, business‑agnostic components (configuration service, queue, registry, SQL/NoSQL storage) into reusable infrastructure services running on a private cloud platform (cloud hosts, storage, virtual networking, load balancers).

Business‑specific components such as user center, payment, search, recommendation, risk control, and geo‑location services were also abstracted into shared services, while the front‑end layer handled request routing, content filtering, and CDN caching.

Business Architecture Optimization

Meituan identified four principles to tame the growing complexity of its business processes:

Simplify complex tasks – remove unnecessary steps and break large workflows into smaller, manageable units.

Standardize simple tasks – define clear specifications, success criteria, and measurement methods for each unit.

Workflow‑ify standards – codify standards into repeatable procedures and manuals for new staff.

Automate workflows – let computers execute the standardized procedures, reducing manual effort.

Applying these principles, Meituan streamlined the “order‑on‑boarding” process. Previously, a contract required manual review, editing, photography, watermarking, and a 7‑10 day turnaround. By digitizing contracts, structuring data fields, and automating validation, the new flow reduced human steps, cut processing time dramatically, and lowered per‑order cost to single‑digit RMB.

O2O Technology Integration

Meituan demonstrated that technology can bridge online and offline operations. For merchants, a cloud‑connected printer receives order details directly from the backend, eliminating phone calls and manual transcription. For users, data‑driven profiling (gender, age, location, device, browsing history) feeds machine‑learning models (e.g., SVM) that predict conversion likelihood, enabling targeted incentives that cut acquisition cost by 35 % and overall spend by 30 %.

Conclusion

The architecture continuously evolves to match business needs: start simple, adopt open‑source components (MySQL, Apache, Lucene, Solr, OpenStack), extract common services, and apply the four optimization principles. By treating both online and offline processes as technical problems, Meituan achieves higher efficiency, lower costs, and a competitive edge in the O2O market.