NIO vs BIO performance
Hi everyone,
Sorry if the answer to this question is obvious to you, but I could do with having my sanity backed up here! I've read a lot about NIO being faster, more scalable, generally more wonderful than traditional stream based I/O, so I've implemented both in the project I'm currently working on.
It's a simple client/server model (one of each) which essentially opens a connection, transfers a load of data and closes it again. The surprising thing (for me) is that the traditional blocking approach runs approx 4 times faster (15secs vs 1m7secs), with both client and server running in separate instances of the VM on my local machine.
Is this just because NIO is designed for high performance when handling many connections from multiple clients (whereas I have one connection from one client) or am I falling into some kind of NIO newbie trap?
Thanks to anyone who can offer advice!
