We are testing a custom Oracle application (Weblogic 11G) on both a Suse Linux 10 server and a Windows 2008 R2 server. Both will be hosted on an HP DL380 G7 with quad processors and 64 GB RAM.
We are limited to free and built-in performance tools for each system. What factors/elements should we be comparing to determine which OS delivers the best performance for the application? I know that memory use and CPU use are two possible elements we can compare. I also found a reference that talks about monitoring the HEAP performance -though I am not totally sure what that is. Any suggestions of realistic elements that we can monitor and measure would be appreciated.
The application performance will be a factor in whether we select Linux as our application server host or continue to use the next version of Windows, since the old application "worked just fine on Windows 2003."