Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

It's all too abstract for me.

Let's assume typical spring microservice - controller layer, service layer, perhaps some persistence or integration with other microservices.

What parts of such microservice will become harder to maintain compared to Loom variant?



Have you looked at the maintainability nightmares that are the reactive model and applications based upon RxJava?

Those types of applications are leagues harder to maintain compared to a virtual thread model.


If you're already writing all callouts from your controllers in a blocking manner then nothing. Because async IO is a pain in the rear lots of libraries try to hide it from you in various ways. However if you're using e.g. pure Java JDBC / SQL drivers, then you should be able to get higher throughput from your web servers and can maybe shrink your server footprint.

However, lots of people don't write code that is just HTTP servers ("microservices"). Also, consider that when it gets easier to write server frameworks there'll be more competition and they'll move faster. So maybe Spring Boot won't seem so attractive in future as someone will have developed something better.


If you're using libraries for everything, nothing. If you need to implement network IO or something, you no longer need to use NIO to get NIO performance. You can just do blocking reads and writes in your threadlet and get the same performance with entirely serial code execution.


But that just means it doesn't really bring that much for 99% of projects. The only time I was writing NIO code was during some Java certification ...




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: