using native device or emcpower device in a multipathed FC-SW fabric

you should always use emcpower devices. if there are any reference to internal path, and that path fails, then the os or powerpath will never be able to redirect i/o on that path to other path.

if you use emcpower devices, powerpath will do the failover and keep track of active and failed paths for each powerpath device.

Thanks Kiran,
that's exactly what I wanted to hear
regards

Alick

Technically for Solaris the correct answer is "it doesn't matter" though in my opinion switching to the powerpath device names is the better habit to get into because for some OS environments it does matter and is a requirement.

To get the PowerPath Product Guide and search on the keyword "Native Devices" and you'll find a page or two explaining how this works on the different operating systems such as Windows, AIX, Linux, etc.

Here is an excerpt from the PowerPath Product guide:

You need not reconfigure applications to use native devices; you simply use the existing disk devices created by the operating system. When using native devices, PowerPath is transparent to applications. PowerPath maintains the correspondence between an individual native device and the path set to which it belongs.

On Solaris and Linux, you have the option of either using native devices (with no conversion of applications) or converting to pseudo devices.