The Problem with Paypal's Node vs Java Benchmarks

You are actually trying to bench mark asynchronous vs synchronous code - and the result is obvious. More relevant may be pitting node against vert.x for example - and node should not be faster than vert.x for most cases at least.