Alibaba Cloud Enterprise Architecture Behind Double 11: A Deep Dive into Scalable Cloud Computing

In 2015, Alibaba's Double 11 shopping festival generated 912.17 billion CNY in sales, setting world records with 140,000 orders per second and 86,000 payments per second, showcasing China's leading computing power.

Alibaba Cloud President Hu Xiaoming highlighted that the computing capability behind Double 11 can be packaged as standardized cloud products and services, allowing any enterprise to build trillion‑level applications and meet extreme business challenges.

Senior Alibaba Cloud expert Shen Xun later disclosed the enterprise‑grade internet architecture—dubbed the "men behind Double 11"—that powered the massive transaction volume and is now being adopted by large enterprises such as Sinopec for digital transformation.

The architecture is organized into five layers: business‑logic presentation, business‑capability operation, cloudified business‑capability, enterprise internet‑architecture capability, and Alibaba Cloud infrastructure.

The cloudified business‑capability layer is pivotal; shared capability centers (e.g., product, transaction, review) are reused across new applications, and dedicated professional teams manage each center, dramatically reducing coordination costs and boosting development efficiency.

Over the past five years, Alibaba has service‑ified all internal systems, exposing hundreds of thousands of unified service interfaces through its capability platform.

Realizing that traditional architectures could not meet internet‑scale demands, Alibaba Cloud developed its own distributed application service framework (EDAS), distributed relational database service (DRDS), and Message Queue to handle rapid development, massive users, large data volumes, and low latency.

After years of iteration, this architecture has proven mature and resilient, consistently surviving the pressure of each Double 11 event.

The core value of Alibaba Cloud's enterprise architecture lies in linear performance growth, exponential reliability improvement, and stable operational costs as the number of machines increases.

Five key characteristics define the architecture: de‑service, decentralization, asynchrony, high availability, and data‑driven operations.

EDAS enables service‑oriented transformation, shortening development cycles and allowing IT systems to accelerate business growth.

Decentralized design eliminates a single traffic hub, reducing load and fault impact, though it raises operational complexity; Alibaba mitigates this with automated, data‑driven operations built from years of incident experience.

Most business processes rely on eventual consistency via Message Queue; for example, Double 11 transaction chains involving over 200 services can be reduced from a 2‑second latency to 30 ms, cutting user loss rates dramatically.

In the cloud environment, thousands of virtual machines support the workload, and automated solutions like DRDS and Message Queue enable second‑level failover and rapid recovery with minimal impact.

The architecture’s data‑driven operation layer helps pinpoint system issues and propose improvements, further enhancing reliability.

Beyond Alibaba’s own platforms, enterprises such as Sinopec, Redridge Capital, and CRM providers have leveraged this architecture to achieve decentralized, asynchronous systems with linear performance and cost efficiency.

Alibaba Cloud continues to open its architecture to more large enterprises, enabling them to build distributed applications comparable to Taobao and accelerate business innovation while maximizing resource efficiency.

Source: Alibaba Cloud