As for the VNAs described here, I'd rather see sample measurements than opinions. With a description of what calibrators, test leads, etc. were used.
One interesting test would be to measure a DUT (from a cold start) with an old stored calibration using the same test fixture.
Looking back through my recent files, I measured an HP 8120-4781 cable back on the 26th July with my old E5071 VNA. This would have been calibrated using an unknown through cal after the analyser was allowed plenty of time to warm up.
Today is the 4th August and I fired up the VNA and loaded up the 8 day old cal corrections and measured the cable again today from a cold start. As soon as the VNA would let me, I took a plot of the cable. This can be compared with the original measurement out to 3GHz.
There's bound to be some phase shift even though I used the same decent HP cal cable (the same way around) as last time but I would expect the amplitude agreement to be really close even though the cal was done 8 days ago and I'm also comparing cold to the original cal 8 days ago (note: the analyser was fully warmed up 8 days ago when the calibration was done).
There's about an hour between the cold and the warm test today. The Cal plot is the 8 day old data for the same cable.
I used a decent HP N cable as part of the initial calibration. This cable is in very good condition and gives repeatable results up to 3GHz although some minor differences are inevitable across 8 days and through removing and reattaching the cable. The main aim is to show the change in drift from cold to warmed up (today) to show how stable the VNA is from a cold start onwards.
Sadly, I did knock/move the cable between cold and warm so I'd really like to repeat this again and make sure I don't touch anything.
To clarify, the warm test was taken about 1 hour after today's initial cold test. The Cal plot is the original from 8 days ago.
The results below are quite good I think. There is hardly any drift at all between all three cases. The E5071 is a very old VNA so I'd expect any modern alternative to be able to compete well against this old workhorse.