I've used my Sun X1, RedHat and windows XP with JMeter + jdk1.4.2. I see no performance difference worth noting. the only real reason to use unix is to be able to ssh/telnet into a system easily and start jmeter in non-gui mode for distributed testing.
the I/O performance between OS is not significant. if I/O is important to you, I would recommend getting a gigabit ethernet, or put multiple ethernet ports on the server and client systems. In general, I/O is a significant bottleneck for stress/load testing. To give you an example, my X1 and linux servers both have dual ethernet ports. If I use the same ethernet port for jdbc queries and stress testing (i.e. http requests), the max through from JMeter's perspective is basically 50% compared to running the jdbc queries through a second ethernet port and switch. In practice, using multiple ethernet ports and switches does provide significant increase in sustained throughput. For burst traffic, there's not much difference. I hope that helps. peter On Thu, 08 Jul 2004 15:56:53 +0200, Ivan Rancati <[EMAIL PROTECTED]> wrote: > hello, > > I was wondering if, on equivalent hardware, there might be a difference > in using JMeter from a Windows 2000 or Linux RH9 client. > Does anybody have experiences to share? > This is only about Java VM performance and network throughput, I don't > mean to start a "Windows vs Linux" war on this mailing list. > > btw, thanks to the authors for an excellent testing tool and prompt support > > Ivan > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
