* P Kishor (punk.kish@gmail.com) wrote:
> Three. At least, in my case, the overhead is too much. My data are
> single bytes, but the smallest data type in Pg is smallint (2 bytes).
> That, plus the per row overhead adds to a fair amount of overhead.
My first reaction to this would be- have you considered aggregating the
data before putting it into the database in such a way that you put more
than 1 byte of data on each row..? That could possibly reduce the
number of rows you have by quite a bit and also reduce the impact of the
per-tuple overhead in PG..
Thanks,
Stephen