Baidu Waimai’s Cloud Migration, AI Logistics, and Architecture – QCon 2017

Cloud Migration from IDC to Cloud

Over the course of one year, Baidu Waimai completed a full migration of its services from on‑premise IDC to cloud infrastructure after the company’s spin‑off. The migration involved more than 20 existing operation platforms becoming unavailable, network connectivity issues between IDC and cloud, and the need to switch hundreds of tightly coupled business modules without any service interruption.

To overcome these challenges, the team built twelve custom operation platforms—including an authorization system, deployment system, network monitoring, CMDB, and intelligent monitoring—using a BVROUTER approach to bridge IDC and cloud networks. They designed a rapid business/data cut‑over solution that achieved second‑level data switching, zero order loss, and zero user perception during the transition.

AI‑Driven Logistics Scheduling

Since early 2015, Baidu Waimai has developed the industry’s first fully automated intelligent scheduling system for food‑delivery logistics. By leveraging massive historical delivery data, the team incorporated machine‑learning and deep‑learning techniques to continuously optimize order assignment, dramatically improving dispatch efficiency, delivery speed, and user experience compared with manual scheduling.

The presentation covered the core design of the intelligent logistics system and how big data and AI are used to enable automatic learning from new scheduling data. It also introduced surrounding subsystems: a “time‑machine” visual tool for reviewing historical dispatch outcomes, a real‑time monitoring and alerting platform for global delivery status, and a simulation framework for evaluating different scheduling models and scenarios.

Scalable Architecture for Rapid Business Growth

Launched on May 20, 2014, Baidu Waimai’s platform has undergone exponential growth, requiring frequent and substantial architectural changes to keep pace with business demands. From 2015 onward, the team performed numerous optimizations across core components such as merchant services and transaction processing.

The talk highlighted the most critical architectural evolutions, including the adoption of industry‑leading solutions, the implementation of innovative designs to solve Baidu Waimai‑specific problems, and the integration of best‑practice architectures that have been successfully deployed in the delivery domain.