Is Spark Still the Best Option for Distributed ETL in the JVM Ecosystem?

I get the interest in Flink, but just a heads up—if you're looking for something lightweight, Flink can feel a bit heavy as it's more of a platform than just a library. If you're aiming for something simple, maybe explore smaller libraries that align more with what you're envisioning.

Yeah, that thread was a bit messy. I think a lot of the confusion came from the wording of your question. People probably misinterpreted it as you saying Spark isn't part of the Java ecosystem when that wasn't your point. Really, it all boils down to your specific requirements. If Spark doesn't meet your needs, looking for alternatives is definitely a good route to take!

Have you checked out Apache Flink? We used it at work a couple of years ago, and we actually preferred it over Spark. It might be worth looking into for your ETL needs!