Building a Stable, Flexible, and Efficient Mobile Operations Configuration Platform – Insights from Meituan‑Dianping

Start‑up companies often begin with rough systems, but once daily active users reach a certain scale, technical leaders must plan systematic refactoring, and revamping the operations configuration platform becomes critical.

Mobile development is becoming increasingly complex, demanding dynamic C‑end configurations such as images, texts, and links. Traditional solutions cannot keep up with rapid version iteration, so Meituan‑Dianping built a stable, flexible, and efficient operations configuration platform, as presented by senior architect Jiang Guobao at QCon Shanghai.

Three key requirements: 1) Efficiency – the platform must be user‑friendly for product teams, avoiding direct database or SQL manipulation. 2) Flexibility – it must adapt to constantly changing product requirements. 3) Stability – from a backend perspective, the service must remain available even under failure, ensuring data can still be retrieved.

How to build an efficient tool platform: - Provide a WYSIWYG visual interface that allows modular page composition, enabling operators to add, delete, and reorder modules easily. - Implement a workflow‑controlled release process with an audit mechanism, separating backend, workflow, and online data to prevent accidental production changes. - Offer a preview/test mode where operators can add their device ID to a test user set and see the post‑release effect before publishing.

Ensuring flexibility during rapid iteration: Store configuration data in JSON format, allowing easy addition or removal of key/value pairs. The UI presents a dynamic page composed of data and control fields, which the SDK fetches and renders on the C‑end.

Achieving stable and reliable configuration data services: The architecture evolved through three stages:

Classic architecture: C‑end → business layer → config service → Redis cache → database. Issues included network latency and Redis instability.

Second stage: added local cache inside the config service to reduce DB latency, but the config service remained a single point of failure.

Third stage: moved caching to the business layer via an SDK, distributing storage and eliminating the central bottleneck, resulting in lower latency and higher resilience.

The platform’s methodology—JSON‑based data, workflow‑driven operations, and SDK‑based distribution—addresses flexibility, efficiency, and stability, helping product, operations, and development teams accelerate C‑end development and iteration.

Author: Jiang Guobao, Senior Backend Architect at Meituan‑Dianping, responsible for the mobile operations platform.