Delivering Always-On Financial Services with
summary
KFINTECH replaces a vertical RDBMS with YugabyteDB and an on‑prem multi‑AZ pipeline (AWS DMS → S3 → EMR/Spark → Yugabyte plus a .NET direct writer), migrating core workloads including 330 million investor folios. The production six‑node cluster (about 192 vCPU) sustains up to 1,000 concurrent connections issuing 300‑upsert batches, with observed I/O ~10k read / 40k write and response times dropping from seconds to milliseconds. Result: zero‑downtime upgrades, active‑active resilience and lower TCO.
AI-written summary · about
chapters (16)
- 0:00KFintech and YugabyteDB intro
- 3:14Mission critical workloads
- 6:47Legacy DB challenges
- 8:53Modern architecture needs
- 12:27KFIN’s data platform evolution
- 16:46YugabyteDB config and workloads
- 19:57Scalability and performance impact
- 20:41Ops simplification with YugabyteDB
- 21:41Seamless upgrades with YBA
- 23:06Built-in observability features
- 24:10Resilience with distributed arch
- 24:52Cost efficiency benefits
- 25:37YugabyteDB internals deep dive
- 26:39Smart driver connectivity
- 27:42Observability and lifecycle mgmt
- 28:05Conclusion
related talks
18mScaling Without Limits: KFIN's Journey to Distributed SQL
25mHow Sharekhan Scaled Trading Platform with Yugabyte's Distributed SQL
57mNo Downtime, No Limits
22mHow Xignite Delivers Real Time Financial Data at Scale
22mMigrating from Monolithic to Cloud Native Distributed SQL Database with Zero Downtime
17mBuilding a Scalable and Performant SaaS Platform with YugabyteDB
26mA Winning Multi Cloud Data Strategy
23mUsing a Modern Database with ML to Prevent Financial Fraud