1. This is not really an issue that should overly concern you
2. This should concern you. If I do feel compelled to create a pagefile partition, I typically make mine 12GB (so that a 4GB page file uses about 33% of the space, which lets a disk defragmenter do its job, if you have an automatically scheduled one. Though defragmentation isn't normally necessary. Many of the performance metrics have suggested that pagefile access is random rather than sequential. Though perhaps your system and work loads behave differently?
Generally the setting you should be using for your page file is
System Managed Size. Explicitly setting a pagefile size is a throw-back to the NT 3.x and NT 4.0 days. Like you wrote, if the page file needs to expand
and it can't expand you set yourself up for problems. If you are really concerned about the performance impact during pagefile expansion, you should really be purchasing more
RAM rather than be concerned about pagefile tuning. I've just never found it to be worth the effort to highly tune it either on a secretary's office PC or on a high-end CAD workstation. It just seems to have little appreciable effect anymore (assuming you have a sufficient pagefile!), however you are welcome to try. I would suggest leaving sufficient pagefile functionality on the system/boot partition so that a proper memory.dmp file from a STOP error can be generated and later debugged by someone.
See also (link)