Integrating Loom in Quarkus | DevNation Tech Talk

Speakers: Arthur Navarro, Clement Escoffier
Loom is among the most highly anticipated projects in the Java world. It promises to address concurrency and Java execution model issues by providing virtual threads. Thus, there is no need to write concurrent programs using asynchronous or reactive APIs; it will be possible to use the traditional imperative model and let Loom handle the rest. The JVM will execute the program and leverage non-blocking APIs automatically!

Sounds good, doesn't it? How does it work, though? Are there any hidden costs? What is Loom going to change in modern Java frameworks? We will answer these questions in this talk. Starting with the integration of Loom in Quarkus, we will compare the different approaches we considered, discuss their respective pros and cons, and show how Loom might change the Java world.

[DmitryShultz]

In the current setup 3500 req/sec seem to be the throughput where Reactive clearly outperforms Loom, would be cool to see the change (if there is any) in the Loom performance if more container resources are added (1GB of RAM instead of 500MB and 1 CPU instead of 0.5 for ex).