No power, no data
Caygill,
The solution must be based in:
"Fault tolerance and graceful degradation".
In this case the idea would be allow a "minimum of information" to allow investigators understand the facts until the sea surface impact.
How it would be designed? We need to prepare an Specification on "what we need".
Obviously in an extreme "electrical power loss" in the a/c systems you may ask: Why record?
Unless you distribute also some sensors.
There is an option for the SSFDR, the accelerometer. That could work integrated to distributed sensors, and that could be made "power independent" of aircraft power during some few minutes.
This is a "System Issue" and IMO there is room for improvement.
What in my opinion is not acceptable is:
Spend years to start (to try) understand what led to this loss.