frequency stability of PLL

I have built a 9 Ghz PLL using a 10 Mhz 0.5 ppm TXCO.
My questions are:

1. Is the final frequency stability 9 Ghz +- 4500 Hz. (0.5ppm) Has the ppm of the VCO no effect?
2. Is there a technique to improve the frequency stability furthermore.(Better than the reference?)

1) it is the same as the reference, since it is locked to the reference
2) tell us more about the system. For instance in a double conversion system, you can design the frequency plan such that the offset caused by LO1 is partially cancelled by the offset of LO2.

How could you discriminate Frequency Stability and Phase Noise at that frequency?
Because I have found that deviation is very low...It might be Phase Noise instead of Frequency Stability..
Be sure that..

there are only 2 ways to improve on stability:
1) get a more stable source, like rubidium
2) design your system so that it self-calibrates the frequency error

Another possibility might be to use some GPS-assisted 10MHz source, like this one:
http://www.tactron.de/images/stories...T/GPS10e-B.pdf