Best Practices

Best Practices

This topic lists the best practices for the optimal use of the Nytro Predictor software tool.

*

Long duration of trace collection could consume more disk space; therefore, avoid using your root filesystem.

NOTE  You might have to run the Nytro Profiler for a reasonable period of time based on your workload to get a better analysis report, for at least 15 to 30 minutes.

*

Run the Nytro Profiler in the production machine to capture the I/O traces in a non-intrusive manner, and then run the Nytro Predictor GUI on a remote machine for analysis.

*

Because the Nytro Profiler generates new trace files, create an empty folder and run the Nytro Profiler from there to make sure that the files are not scattered across different locations.

*

Because of the inherent blktrace issues, do not run two instances of the Nytro Profiler tool to capture the traces on the same device.

*

On a Linux machine, you must run the Nytro Profiler on a session that does not time out.

*

If the Nytro Profiler is running on a heavily loaded target machine, use the CPU command line option to reduce the CPU usage of the Nytro Profiler application.

*

Copy the I/O trace files onto disks that are not being used by the Nytro Profiler.

*

Do not copy the I/O trace files onto the operating system disks.

*

Keep track of the disk usage consumed by the trace file during trace capture.

For more information, see the Nytro Predictor Tool Usage topic.