Hi Jan,

Its relative frequency if you've done a relative pnoise sweep setup i.e. 1M in your case. For absolute sweep its the absolute frequency so 4.001G.
If you're unsure and you did a pnoise sweep, its easy to get the form to print spot noise at a couple of frequencies, from that it'll be easy to understand. For example if i did a pnoise sweep 1K~100K relative to a 4G carrier, if i print noise summary at 1K and 100K, i'll get different numbers. If i print at any number >100K i'll get the same result as at 100K. So I know the relative frequency is used for the form in this case.

As to which is important; spot noise or integrated noise, it depends on your system (or specification). For a example a receiver which will have a large blocker at a certain offset frequency, the spot phase noise at that offset would need to be known for one to calculate the reciprical mixing product that would affect your receiver.
An example for when intergrated noise would be relevant would be if this VCO signal was used to upconvert in a transmitter a frequency modulated signal, in such cases the integrated phase noise within the modulation bandwidth would create residual FM which would degrade the transmitted signal SNR... degrading the ability to demodulate it at the receive end.
In most well defined/designed systems, you'll need to meet phase noise at at least a couple of spot frequencies (from which an integrated phase noise number is easily estimated), where ever it is the limiting factor to a particular performance metric of system.

 
cheers
aw