How to write fast in Apache Ignite
summary
The talk compares three different Apache Ignite APIs for writing data to the cache and shows that the recommended API is not always the fastest. It explains pitfalls, demonstrates several performance tricks, and presents benchmarks that quantify how factors like enabling SQL, cluster size, and replica count affect write throughput.
AI-written summary · about
chapters (21)
- 0:00Introduction
- 2:24Fast writes in Ignite
- 5:30First write attempt and results
- 6:11Data Streamer API limits
- 9:28Replication impact on speed
- 11:14Write mechanism and implementation
- 13:48Batch writes and benefits
- 17:25Fixing old log issues
- 17:46SQL write speed impact
- 19:30Data size dependency
- 20:11Optimal batch size
- 20:33Replica count impact
- 21:59Client types and features
- 24:09Server-side code execution
- 25:53Server-side Data Streamer
- 27:01Binary Object creation
- 29:49Binary Object creation comparison
- 31:19Parallel prep and write
- 33:50Entry Processor for updates
- 36:19Summary and recommendations
- 37:26Q&A
related talks
59mApache Ignite Persistence
42mEfficient File Handling for Java Developers
44mHow to speed up your application using JCache
01h 02mHow We Used GridGain to Build a Telephony API System
50mA Java Implementer's Guide to Better Apache Spark Performance
47mHunting issues in a Java app with 31G heap and 500G off-heap: an Apache Ignite case study
45mCaching in Applications Still Matters
51mJava at Speed: Getting the Most Out of Modern Hardware