Java, Python, and even Ruby [1] can get you under 10ms for service responses that actually hit a database, in my experience. But why improve performance by switching to a language that's well suited to the task at hand when you can do it by switching to a language that's hip? [1] My company ran into a problem where one of our Ruby service responses was taking something about 100ms. We ruled out the database, and the…
With some work you can write some Java that will easily take a message, parse it, do logic, and reply in under .1ms. Not that useful for a webapp though, databases are kinda cool.
With Java you can receive an order, parse it, submit it to your trading engine, persist it, match it, and return the response in less than 4ms