[erlang-questions] optimizing mnesia record size
Fri Apr 11 23:22:05 CEST 2008
i'm trying to decrease the memory footprint of an mnesia database. i
turned some strings into binaries and got some space savings. past this,
i'm sure how to proceed.
one thing that would help is, is there a way to compute what the record
size will be (maybe within a constant factor?). i've been playing around
at the shell calling size/1 on erlang:term_to_binary/1 but i'm not even
sure this is indicative.
also, are there some guidelines? i was a bit surprised to discover that
replacing an atom with binary_to_list/1 composed with list_to_atom/1
actually made things bigger (by previous paragraph's metric). i was also
surprised to discover that really short strings are actually better as
lists rather than binaries.
any pointers would be greatly appreciated.
In an artificial world, only extremists live naturally.
-- Paul Graham
More information about the erlang-questions