Scaling Innovation: Rakuten’s Cloud-Native & AI Journey with
summary
Rakuten runs Yugabyte across Kubernetes on three active sites (Tokyo, Osaka, Kobe) to support peta‑byte scale OSS and AI workloads, ingesting via Kafka into a single backend that serves both YSQL and YCQL apps (KYC, fraud, time‑series metrics). Claims ingest of ~12 million million records in five minutes, achieves active-active three-site geo-redundancy, performs a zero-downtime 30-day cluster upgrade, and reports 200% faster read/write after switching encryption—enabling real‑time ML for sleeping‑cell detection and automated operations.
AI-written summary · about
chapters (9)
- 0:00Rakuten’s cloud-native AI journey
- 1:04Why distributed SQL over legacy
- 3:09Yugabyte’s SQL and NoSQL APIs
- 9:36Yugabyte’s resilience value
- 12:44Edge to AI insights journey
- 15:35Cultural and technical hurdles
- 19:27Zero-downtime upgrades with Yugabyte
- 20:52Future AI features
- 23:19Closing remarks
related talks
30mFrom monolith to petabyte scale: Built a cloud-native, AI-ready data layer
26mCloud Native Database Meetup Japan: The State of Distributed Database in Japan
26mLeveraging a Distributed SQL Database at Tokopedia: Best Practices & Lessons Learned
25mHow Sharekhan Scaled Trading Platform with Yugabyte's Distributed SQL
30mBuilding Cloud Native Applications on Distributed SQL Platform
36mFireside Chat with Kroger: Examining their Two-Year Distributed SQL Journey & What's Next
34mWhat’s New: Discover the Latest Innovations and Updates from Yugabyte
57mNo Downtime, No Limits