I've created two protection groups in my SRM (6.5.1). The protection groups contain separate datastore groups and separate virtual machines. I've created two different recovery plans. Each recovery plan has one of the protection groups in it so that I could fail them over separately. I presume that would be okay correct? I could failover one protection group to the recovery site while the other protection group remains running at the protected site without impact. I'd assume this is possible but wanted a sanity check before I attempted it.
If you want to sanity check your recovery plans you can execute Test Recovery and check that everything is correct. You can find more information in documentation - Differences Between Testing and Running a Recovery Plan.
Hope this helps,
VM's associate to those protection group is only impacted. So when you created the recovery plan and tagged a respective protection group any VM's under that PG will failover when you run a recovery plan.
There are 3 different ways of initiating a recovery plan
Test recovery is used on a regular basis to validate the VM recovery and to perform a complete failover Planned migration is executed if required downtime is available for production servers part SRM protection group.
Hope that clears.