Building a Unified Data Analytics Platform at TCL Using StarRocks

Background – TCL, a multinational group with over 130,000 employees, is undergoing its fourth major digital transformation, aiming to standardize business processes and consolidate data across its subsidiaries. To support this, TCL is building a shared big‑data platform that integrates data from various business units.

OLAP Construction Journey – Historically, TCL used relational databases and Hive‑based data warehouses on Hadoop. As data volume grew, ClickHouse was introduced for high‑performance analytics, but increasing concurrency and real‑time requirements led to the adoption of StarRocks, an MPP database that simplifies operations and improves multi‑table joins and write performance.

Typical StarRocks Application Scenarios

1. Real‑time Dashboard (New Ark) – StarRocks powers a real‑time data warehouse that feeds a marketing‑centered dashboard, handling both high‑frequency and batch data to support large‑screen visualizations and self‑service BI.

2. Group HR Service – By consolidating HR data from multiple E‑HR systems, StarRocks enables daily and hourly snapshots, reducing query latency from ~15 seconds to 3‑5 seconds for complex multi‑table joins.

3. Email Alert System – StarRocks handles massive write and read workloads for security‑related email alerts, delivering sub‑second response times and outperforming previous Kudu‑Impala solutions.

Future Plans – TCL intends to further integrate StarRocks with its proprietary big‑data platform, migrate more OLAP workloads to StarRocks, simplify the technology stack, achieve unified real‑time analytics, and improve reliability through enhanced monitoring.

Conclusion – The sharing session highlights TCL’s practical experience with StarRocks, demonstrating its suitability for large‑scale, high‑concurrency analytics in an enterprise environment.