Optimization Practices of Meipai Live Streaming Architecture

Meipai Live Streaming Architecture Optimization Practice

In live‑streaming scenarios, the transmission path from the anchor’s device to the audience’s screen is extremely long, and every link directly affects the viewing experience. Since its launch at the end of 2015, Meipai Live has undergone two years of evolution.

During this period, the team built a full‑chain monitoring system, optimized the push‑stream SDK and the player, and adopted a multi‑source, multi‑CDN architecture. These efforts continuously improved the stability and quality of the service. The current direction is moving toward intelligent scheduling based on live‑streaming data analysis.

The presentation shares several practical attempts made by Meipai to enhance experience and availability, including performance tuning, container‑based platform construction, cloud storage services, live‑streaming services, and instant‑messaging services.

Speaker

Ma Junsen – Technical Director of the Architecture Platform Department at Meitu (Shenzhen). He is responsible for performance optimization, container platforms, cloud storage, live‑streaming services, and IM services. Previously, he worked at Sina Weibo and Qihoo 360 as a technical expert, handling multi‑IDC architecture, IM services, RPC Motan, cache services, KV storage, and foundational frameworks.

The PPT slides are available for download. Extraction code: z403 .