4 ms·

Not only it is easy to use, but the source code is a real pleasure to read, contrary to Hadoop's mess.