Case Study: DMM.com achieves real-time, high-accuracy product recommendations with Cloudera Enterprise

Provide Recommendations to Online Users with Big Data Analysis of the Infrastructure That Uses Apache Spark

DMM.com Labo, the systems and web‑marketing arm of the diverse DMM.com group (video streaming, e‑commerce, games, and more), faced fragmented, division‑specific recommendation and behavior‑analysis tooling and slow, hard‑to‑tune outsourcing. The company needed a unified, real‑time analytics infrastructure to improve prediction accuracy, reuse models across services, and enable faster in‑house customization.

They built an in‑house big‑data platform using Cloudera Enterprise with Apache Spark and Kafka on Intel® Xeon® E5‑based, SSD‑backed servers to support in‑memory, parallel processing. Deployed in late 2014 and rolled out group‑wide in September 2015, the system delivers high‑accuracy cross‑service product recommendations from purchase histories, real‑time behavior analysis for business intelligence, and self‑service SQL access for marketers, with ongoing work to scale and add streaming trend analysis.

DMM.com

Norio Ojima

CTO, Office Chief