How to Use the new JDBC 4.3 Sharding API for Massive OLTP Scaling
summary
Claims that JDK 9’s JDBC 4.3 sharding API lets clients pass sharding keys and ConnectionBuilders so connection routing eliminates extra network hops; demos use YCSB and SwingBench against Oracle Sharding and Oracle TimesTen Scaleout. Measured results: plain JDBC across 16 nodes ~2M ops/s, JDBC+sharding ~5M ops/s, TimesTen scaleout shows tens of millions (38M) for mixed workloads and up to 1.4B reads/s for read-only tests. The distributed DB is built and running in production.
AI-written summary · about
chapters (39)
- 0:00Introduction
- 0:17Job Titles
- 0:59Database Scalability
- 1:31What is Scalability
- 2:19The Scalability Boss
- 2:57Scalability Curves
- 3:55Performance Curves
- 4:55Vertical Database Scaling
- 6:04Scaling Limitations
- 7:25SGI Camp Ring
- 7:48Hardware
- 9:58Scaling horizontally
- 11:19The fundamental issues
- 11:42No sequel database
- 13:15What is the Sharding API
- 13:26Sharding Databases
- 14:35Benefits
- 15:32Benchmark
- 16:10CSP Benchmark
- 17:35Comparing NoSQL Databases
- 24:10Where to use
- 25:04Westies Benchmark
- 27:01YCSB
- 27:33Why is it so big
- 28:55How do we do it
- 30:03We used another memory database
- 31:03We used the Oracle API
- 31:42Keep your data model
- 32:03Special syntax
- 32:25What does this database do
- 34:04What about high availability
- 35:00Replication failures
- 36:14Worlds fastest OLTP database
- 36:28China Mobile
- 40:46Linux
- 41:56Cloud
- 42:40Terraform
- 44:07Recap
- 45:34Download
related talks
32mSharding for mission-critical, internet-scale apps
23mSharding for hyperscale and geographically distributed relational databases
46mDatabase Sharding the Right Way Easy, Reliable, and Open source
27mScaling for Intelligence: Distributed SQL Databases in the Age of AI
52mTo Shard or Not to Shard
29mOracle TimesTen: extreme performance for latency-critical applications
27mAutomatically shard & scale-out traditional databases on K8s
04mKuassi Mensah Interview