Inside a plasma globe
The high voltage inverter is a very simple circuit, using classic parts (CD4069-based oscillator, TIP122 power transistor). The transformer ressembles a lot a TV flyback.
The output voltage is about 10kV (according to the length of the arcs I am able to draw, I have no voltmeter which can withstand such high voltages), and its frequency is a dozen kHz. This high frequency allows arcs to appear around just one electrode inside the plasma globe (by capacitive coupling).
Another effect of such an high frequency is that the insulating layer of two powered cables gets covered by a blue glow when the cables are close enough from each other (I was unable to take a picture, the glimmer was too low). This phenomenon is accompanied by a strong smell of ozone and another irritating gas (probably some nitrogen oxyde). The insulation of the cables doesn't look damaged, and I have never seen this with high DC voltage (even with 20-30kV, there is only a brutal breakdown and the insulation gets destroyed). This must be related to http://jlnlabs.imars.com/plasma/html/s_gdp.htm.
It is also possible to obtain an effect similar to the traditional plasma globe in a classical 230V light bulb (sorry for the picture quality, it was taken in darkness).
Streamers look however hotter than in the normal globe (higher gas pressure) and tend to appear always at the same places (where the filament has spikes).