Don't forget Jelastic, very nice, doesn't have any special APIs.
http://jelastic.com/
In a move that's becoming all too familiar, leaders of the effort to develop Java EE 7 – the next version of Oracle's Java platform for enterprise computing – have recommended that certain planned components be deferred to a later version in the interest of keeping the project on schedule. "Despite our best intentions, our …
Nobody wants Java to be a mega-system. The JVM is quite impressive, language is functional, and most of the standard edition libraries work very well. Oracle should stop throwing everything into the bloated "Enterprise Edition" classification and focus on simple modular solutions to specific and well defined problems. They might even make some money on it. A marketing model of creating specifications so complex that only Oracle can provide implementations isn't going to work. As soon as I see a new feature listed for "Java EE" I stop reading.
The article says that Java development plans were overambitious. That sounds strange. Java is stuck on time. The language is very unproductive compared to C#, so much that motivates developments such as Gosu, Scala and Kotlin. Java EE, since early versions, lacks session isolation, which is necessary for many robustness requirements. Java SE API is old and cumbersome. The only barely up-to-date Java product is the JVM, and although it's regarded as a masterpiece of optimization and management, it's considered too heavy for mobile devices and even some PC applications. Java is not "run anywhere" anymore - it doesn't run on iOS and won't run on Windows Phone. And Oracle strategy to sue Google left most Java developers afraid of future (just every Android developers, and most of open-source fan boys).
I think Java plans are not overambitious - they are hindered at development hell. It seems that Java is somewhat cursed - everybody that touches it gets some degree of failure or at least a shrouded future.