In case a system is already up and running (being used) and is being retrospectively validated, how is the IQ for the same to be performed?
Do we need to show an IQ for the same in the first place?[/quote]
My personal experience with IQs has been pretty positive. They (IQs) provide a nice means to document the configuration, ensure the system is set up properly, and provides a solid baseline for any future upgrades / changes. Keep in mind that the IQ/OQ/PQ paradigm is just a well-established convention. If you can justify that your system is validated by a different means, you are free to do so. Keep in mind, though, auditors tend to have a comfort level with what they know. Deviations from the de-facto standard often leads to more extensive investigation.
If there are different instances of the application, one for development, one for testing and validation and the other production instance. Is it required to do an IQ for all the instances or is it sufficient to show that the system is properly installed if the recommended hardware and the installation procedure is followed only for one instance. [/quote]
If you can wrap it all up in a single IQ (maybe covering all the installation differences on the various instances?) then I wouldn’t see a need for separate IQs. Consider documenting your rationale (i.e., in a Validation Plan) for whatever direction you choose. Then, at least, there is solid justification for what you do end up doing.