I don't think it's a fair criticism of Ballista to say that it failed in any way. It just looks to need substantial effort to bring it on par with Spark. The performance benefits are meaningful. Ballista can then not only take the crown from Spark, but also revalidate Rust as a language.