Alibaba’s 2019 Double‑11: Full Cloud Migration and Live‑Streaming Leap

1. From iQIYI CTO to Alibaba Retail Tech Leader

In 2019 Tang Xing (nickname Pingzhou) joined Alibaba Group as Vice President, heading the new retail technology division that includes Taobao, Tmall, Rural Taobao, Xianyu, and other platforms.

Before Alibaba, he was CTO of iQIYI, where he built the world’s first video‑understanding system called the iQIYI Brain and launched the pioneering video‑social product iQIYI Bubble.

During the 2019 Double‑11 event, Tang was appointed the technical leader for the Taobao ecosystem, transitioning from video to e‑commerce while preparing for massive traffic spikes.

Understanding the inner connections and differences between the two fields makes switching and adapting much easier.

2. The 11th Double‑11: A Transformation Comparable to Wireless Migration

In 2019, Double‑11 live streaming became a milestone in Taobao’s content ecosystem, comparable to the wireless shift of 2015.

Taobao’s 2019 Double‑11 live streaming was a transformation as significant as the earlier wireless migration.

The live‑streaming sales reached nearly 200 billion CNY, accounting for over 50% of brand growth, and required stringent demands on messaging channels, networks, and CDN resources.

To meet these demands, the Taobao live‑streaming team built a decentralized real‑time audio‑video communication engine, reducing latency from 5‑7 seconds to under 1 second and improving performance on weak networks.

Practice shows that the most serious issues during Double‑11 are often not the ones we anticipate.

We can never prepare for every problem; unknown challenges remain, and we must continuously upgrade technology to approach the optimal limit.

For technologists, Double‑11 never has a “best” version—only a better one.

3. A Year‑Ahead Decision to Move to the Cloud

This year’s Double‑11, we put the “most critical” systems on the cloud.

In January 2019, Alibaba launched the “Cloud‑Future” project, migrating hundreds of core e‑commerce applications and over a million CPU cores to the cloud.

The migration was staged: Q3 saw large‑scale application development, with infrastructure preparation and early validation during the 618 promotion.

Stable migration to the cloud;

Stable operation of cloud services;

Exploring new, efficient development methods using cloud technologies.

Phase 1: Application migration. Risks stemmed from changes in cloud infrastructure and middleware, prompting upgrades such as middleware library updates, version checks, asynchronous logging, and cache cluster migration, all managed by an automated upgrade detection system.

Rigorous testing, traffic replication, and rapid failover mechanisms were employed, establishing four logical data centers for one‑click disaster avoidance.

Phase 2: Establishing a cloud‑assurance team. A cross‑functional group aligned cloud and e‑commerce development, defined response mechanisms, integrated change and alarm systems, and conducted simulated attacks to improve emergency handling.

Dedicated PMO support ensured regular sync meetings and coordinated emergency procedures across teams.

Phase 3: Leveraging cloud to explore new development approaches. Alibaba Cloud’s bare‑metal servers and ASI containers were widely adopted, and a serverless function platform (FaaS) named Gaia was built to serve multiple business scenarios, dramatically improving development efficiency.

During the migration, infrastructure became richer while development became lighter, shifting from pro‑code to low‑code and freeing productivity.

4. Edge AI Decisions: 14 Billion Inferences on Double‑11

Prior to cloud migration, Alibaba’s retail division had already deployed on‑device AI to recognize user intent in real time, powering personalized recommendations and live‑streaming product detection.

On Double‑11 2019, edge‑AI made 14 billion decisions, dramatically improving recommendation relevance and user experience.

The team also began combining deep learning, computer vision, and 3D technologies for applications in home decoration and furniture.

While AI has made rapid progress, Tang cautions that over‑hyped expectations can lead to perceived bottlenecks; true breakthroughs require deep integration with business, product, and engineering over time.

Successful AI adoption depends on solid infrastructure, data, and product development, not just algorithms.

5. Future Plans

Post‑2019 Double‑11, Alibaba aims to further enhance core checkout experiences, optimize resource usage, and reduce costs through continuous technical upgrades.

Upcoming focus areas include 5G‑driven innovations such as ultra‑low‑latency live streaming, AR interactions, and advanced encoding, as well as expanding AI, edge computing, and physically‑based 3D rendering for intelligent home design.

Interview Guest

Dr. Tang Xing, Vice President of Alibaba Group, leads the new retail technology division covering Taobao, Tmall, Rural Taobao, Xianyu, and more. He previously held senior roles at Google Shanghai and iQIYI, driving video search and social video innovations.