Sriram83 schrieb:
2/. Tatu - we are giving sufficient time for JVM to warm-up. First we run a load test with 1 user for 15 minutes. After this test, we ran with 3 users for 1/2 hour. The times I mentioned below are average time taken for around 1500 transactions.
It's not clear to me how you're doing the testing. Just to make more explicit what Mukul and Tatu have already said: You must set up things so that you have only one JVM startup, and then a hundred (or more) consecutive transformations in that process. If, however, you restarted the JVM over and over again (thus warming up the operating system), you'd *likely* be testing the operating system speed at starting the JVM, not the JVM speed at running the XSL transformation. Finally, as Mukul said, the difference is not spectacular. Maybe the Windows implementation is, horribile dictu, simply a little bit faster at this particular task. If indeed this is so and you can't live with it, you know where to get a Microsoft Server licencse :-) Michael Ludwig
