High Availability and Disaster Recovery Architecture: The Evolution of Alipay’s System Design

High availability and disaster recovery (DR) are critical for enterprise services, cloud computing, and mobile internet platforms, ensuring uninterrupted service and user confidence, especially during traffic spikes such as China’s "Double 11" shopping festival.

Early Alipay architecture (2004‑2011) relied on commercial load balancers and a single database per core system, leading to single‑point failures and limited DR capabilities.

In the second stage (2011‑2012), Alipay split logical data centers, introduced soft load balancing, and implemented horizontal data sharding based on user UID, reducing single‑point bottlenecks and enabling multi‑datacenter active‑active deployment.

A dedicated Failover layer was added to handle master‑slave switchovers within minutes, preserving data integrity and minimizing service disruption.

From 2012‑2015, the team tackled DB connection limits, IDC resource constraints, cross‑datacenter latency, and introduced unit‑based architecture, separating core and non‑core services into distinct units (A, B, C) with localized data and independent traffic control.

Blue‑Green deployment was adopted to limit user impact during releases, using separate Blue and Green groups within each unit and gradually shifting traffic for verification.

The final unit‑based, multi‑active architecture provides flexible traffic control, scalable resource allocation, and rapid disaster recovery across data centers, achieving strong high‑availability and DR capabilities.