How Alibaba Cloud Powered a 268.4 Billion‑Yuan Double 11 Traffic Surge

"Not every cloud can handle this traffic. China has two clouds: Alibaba Cloud and the others," said Alibaba Group CTO Zhang Jianfeng, highlighting that Alibaba Cloud was built from the first line of code a decade ago and runs on a self‑developed cloud operating system called Feitian.

During the 2019 Tmall Double 11 shopping festival, Alibaba achieved a transaction volume of 268.4 billion yuan, a peak order rate of 54.4 k orders per second, and processed 970 PB of data in a single day.

For the first time, Alibaba moved 100% of its core systems to the public cloud, making it the world’s first large‑scale internet company to run its entire core infrastructure on a public cloud platform.

The Feitian operating system, developed in‑house for ten years, underpins the cloud, while the Shenlong server architecture provides linear performance scaling under heavy load.

Alibaba also deployed self‑developed databases: OceanBase, which topped the TPC‑C benchmark globally, and PolarDB, both of which handled the Double 11 workload without issues.

Compute and storage were decoupled, enabling seamless scaling of storage pools, and the company introduced a large‑scale RDMA network—the first of its kind globally—to achieve faster remote storage access than local disk.

These four breakthroughs (Feitian OS, Shenlong servers, self‑developed databases, and RDMA networking) together delivered a massive performance uplift, allowing real‑time processing of up to 25 billion records per second during the event.

Alibaba also operates the world’s largest message‑queue system to manage the massive flow of messages across its services.

Looking ahead, Alibaba unveiled its self‑designed AI chip, Hanguang 800, which will power a large portion of AI workloads in the next Double 11.