How Suning Moved 26,888 Servers in 75 Days – Key Takeaways

Suning, a retail technology enterprise, faced the daunting task of migrating an entire data center. The project required four key steps—organizational planning, resource preparation, migration execution, and post‑migration summarization—and involved over 2,000 participants handling 26,888 servers, roughly a quarter of Suning’s systems.

At the end of 2019, driven by a need to upgrade its data center, the Banqiao migration team committed to completing the move within six months—a first of its scale for Suning. The challenges were multiple: tight timelines due to frequent promotional windows limited migration slots to the early‑morning hours (1 am–6 am); extensive communication was required to map numerous machines and systems; and the operation had to coexist with daily duties while coordinating nearly thirty sub‑project groups.

Project manager Li Zhongqiu recalled the most memorable moments: “System migrations can fail at any time—late night or weekend—so every team must respond instantly. Once, a logistics‑center LPSS migration failed on a Friday evening; we held an emergency call at 8 pm with participants joining from the road, the subway, and elsewhere. That collective effort made us realize we were all in this together.” The migration succeeded thanks to rapid response and tight collaboration across teams.

Technical incidents were also resolved swiftly. When a load‑balancer real_server pool overflow caused an error during capacity expansion, the issue was escalated by technical lead Chen Lianhai, who coordinated with the network management department to provide both temporary and long‑term solutions, eliminating the risk on the same day.

One notable effort involved migrating Sentinel, a core shared component. Nearly 1,000 systems and over 5,000 shards were registered across two Sentinel groups. Zhang Yachang led the migration with the caching team, completing six batches without any downtime.

According to chief technical officer Chen Lianhai, efficiency grew exponentially: the first batch moved 200 groups, the second batch 400, and peak nights saw over 1,000 virtual machines migrated. The team distilled the success into five key practices:

Set clear goals and break them into sub‑goals; execute without retreat.

Prioritize difficult tasks first; start fast, then ease the pace.

Identify risks early and implement mitigation measures; avoid ad‑hoc decisions.

Escalate unresolved issues promptly and seek cross‑team assistance.

Maintain strong teamwork and personal dedication.

This migration was Suning’s first at such scale and finished more than half a month ahead of schedule, providing a valuable case for future data‑center adjustments. System stability improved, migration tools delivered long‑term operational benefits, and both operations and development staff deepened their understanding of architecture and infrastructure, dramatically boosting skill levels.

Cloud‑capacity gains were evident: migration tools for web, app, cache, and database workloads increased operational efficiency by tenfold, IAAS resources were accurately forecasted and pre‑allocated, and annual leasing costs were reduced by over ten million yuan.

The successful migration underpinned the massive traffic surge of the recent 618 Super Show, where 1.2 billion viewers generated over 5 billion yuan in sales, highlighting the critical role of robust infrastructure.

“This year’s 618 will be remembered especially well,” the team concluded.