Hey, Doze! Good to see ya here.
Leaving MA370 to one side, are there any accidents in recent times where the investigation would have significantly benefitted from more CVR time, let alone offsite storage?
You have a point there. But the real issue is not the total time recorded, but rather when the recording stops ( although I feel that 6 hours or more would be better for the long haul flights). A second consideration is the power supply. This was shown in the Swiss Air crash. The electrical failures due to the fire resulted in the failure of the ACARS, CVR and FDR. The jet hit the water a few minutes later. Finally, offline storage, like Carbonite, et al, seems a good idea. At least the investigators would have "something" to look at and analyze. Saw this with AF447, huh?
So with 370, we may have a high-quality, two hour recording of wind noise and such, and then "thump". That's if the recorder was still running.
If the data recorder was still running, then we should have data all the way back to takeoff, if I understand the current protocol for the T7. Not sure if it would have all the ECAM alerts, but need a T7 type to let us know.
Up to me, I would "demand" battery backup and a power supply that would be the last to go in an electrical fire. Fer chrissakes, my wife's iPad can run for 8 or 10 hours easily. Our dinosaur solid state recorder in the Viper ( on the ejection seat) only had a few minutes of data, but it captured several parameters of value for the accident investigation.