From monolith to petabyte scale: Built a cloud-native, AI-ready data layer
summary
Migrates a live telecom data stack from multiple stores into a unified Yugabyte/YSQL platform and runs it at 1.8 petabyte for one customer (projected to double in a year). Architecture is active‑active across three Japan DCs with 72 pods across three DCs (24 cores, 144 GB, 35 TB per pod) and uses Yugabyte Anywhere plus Hexa Cluster for zero‑downtime migration. Observed throughput is 50k–150k ops/s with writes <20 ms, reads <50 ms, updates <200 ms and no user‑facing downtime.
AI-written summary · about
chapters (12)
- 0:00Rakuten Symphony data platform
- 3:22Global business and cloud vision
- 8:44Why Yugabyte chosen
- 13:42Monolith to Yugabyte architecture
- 16:53Active-active across 3 data centers
- 19:21Live migration and production workloads
- 22:13Migration challenges and HexaCluster
- 22:34Petabyte-scale metrics and specs
- 25:44Cloud-agnostic and microservices benefits
- 27:31Feature requests and telecom support
- 28:16Zero-downtime voice layer
- 28:59Acknowledgments and petabyte journey
related talks
26mA Winning Multi Cloud Data Strategy
22mMigrating from Monolithic to Cloud Native Distributed SQL Database with Zero Downtime
35mA Customer’s Journey: Migrating from NoSQL to a Cloud Native Distributed SQL Database
28mDisrupting the Status Quo with Distributed SQL to Deliver Data-Driven Solutions
23mAdmiral Scales Globally with YugabyteDB on GCP + Single-Digit Latency
25mScaling Innovation: Rakuten’s Cloud-Native & AI Journey with
19mEvolve: A Database Journey from Ground to Cloud
30mBuilding Cloud Native Applications on Distributed SQL Platform