How ClickHouse Powered Suning’s Billion‑Tag User Profiles in Seconds

Case Overview

Suning Technology’s senior architect Yang Zhaohui presented a case at the 2020 TOP100 Summit describing how ClickHouse was used to rebuild Suning’s tag platform for massive user profiling.

Background

To support marketing activities, Suning needed to identify target audiences and user characteristics. Their existing tag system covered billions of tags for 600 million users across retail, finance, sports, and other industries. The challenge was to achieve second‑level query performance on this scale.

Solution and Key Points

The team re‑architected the tag platform using ClickHouse, optimizing the storage model, data distribution, and query flow. They integrated RoaringBitmap, a compressed bitmap library, allowing a single row to store massive ID lists efficiently. This leveraged ClickHouse’s distributed parallel computation, delivering billion‑level tag queries in seconds.

ROI Analysis

Query response time reduced from over 20 seconds to 2–3 seconds for billion‑scale user profiles.

Near‑real‑time tag data updates.

Linear scalability of storage, import, and query performance by adding machines.

Hardware resource usage cut by about 50 % compared with an ElasticSearch‑based solution.

Tag expressions and SQL queries are user‑friendly, improving maintainability.

Insights

Choosing a robust open‑source solution and deeply customizing it for specific business scenarios can quickly create value, avoid reinventing the wheel, and contribute enhancements back to the community.