directory-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alessandro Torrisi <>
Subject [mina] Thread Pools, Concurrent connections and Performance
Date Tue, 27 Dec 2005 09:07:25 GMT
Hi ! I'm developing a free and opensource Direct Connect software (P2P 
server). Now that protocol implementation is quite complete I'm testing 
with lot of connections.
When connections are made in a concurrent way (50-100 a time), the 
server seems to be blocked...
Can I do something to improve performance, adjusting some parameter or 
applying some programming pattern directly on Mina ? I've listened about 
Thread Pools on SocketAcceptor and IoThreadPool but I didn't find any 
tutorial or documentation, is it possible to directly configure these ones ?
A C++ hubsoft (for example verlihub) can have over 10.000 concurrent 
connections on a 100 Mbit line, on my personal line (10 Mbit) it can 
have 2000-2500 users. I tested it with 600-700 connections, the software 
hung up.
It could be that my software is not well optimized, but even if I 
doubled the used bandwidth I can't see 1000 users !
I listened about Mina scalability and pointed on it...please help me !
What are the base step to introduce a new more performant architecture 
without rewriting the code from scratch or upset the code ?

Best regards (and excuse me for my english),

View raw message