Unlocking Data-Intensive Applications with Virtual Threads
summary
Claims that Java 21 virtual threads let a web app serve 100K requests/sec by mapping millions of lightweight virtual threads onto a few platform threads, demonstrated with Spring Boot/Tomcat and reactive alternatives. Load tests use JMeter, k6, Predator and a Raspberry Pi cluster for generation; results show huge throughput gains but shifting bottlenecks to JDBC/databases and networks. Examples include Super Bowl–scale numbers (~70k req/s peaks, 4.5B votes) and advice to use Kafka/R2DBC/streaming to handle persistence.
AI-written summary · about
chapters (12)
- 0:00Introduction and speaker background
- 1:47History of Big Data and Java/Scala trends
- 5:21Understanding data-intensive applications
- 10:38Platform threads vs virtual threads
- 12:28How virtual threads improve throughput
- 15:21Project Loom and Java’s coroutine implementation
- 18:09Benefits of virtual threads
- 19:57Testing virtual threads with Spring Boot
- 23:05Database bottlenecks and solutions
- 26:58Future of Java and virtual threads
- 31:16Key takeaways and call to action
- 35:12Q&A session
related talks
46mJava 21: The revolution of virtual threads - A Deep dive
51mJava 21: The revolution of virtual threads
50mGet The Most Out of Virtual Threads in Java 25 -A Deep Dive into the Latest Achievements of Virtual
47mMore Action, More Overview – From 4k to 1M Threads
49mJava Dark Threads
47mVirtual Threads, Structured Concurrency & Scoped Values: Putting it All Together
57mJDK 19 Virtual Threads: Skip Reactive, Boost Throughput
38mJava Project Loom: Why Should I Care?