[erlang-questions] understanding the scaleability limits of erlang and mnesia

[Scott Lystig Fritchie]

If I recall correctly, a single modification to a Mnesia table (or
Mnesia fragment) while a replica is down will trigger the transfer of
the entire table/fragment from an always-alive -> now-recovering nodes.

For the sake of a thought experiment, if the 2GB size limit on a dets
table were removed, and if you stored 2TB of data in a table ...

... then deleting a single 1KB record in that table while a replica was
unavailable would require transferring 2TB of data to resync, plus
whatever would be required to "catch up" with table modifications that
were made during the re-sync.

-Scott