On 2011-01-15, bubba postgres <bubba(dot)postgres(at)gmail(dot)com> wrote:
> --000e0cd332267ac2e60499d7ceb2
> Content-Type: text/plain; charset=ISO-8859-1
>
> I've been googling, but haven't found a good answer to what I should do if I
> want to store time series in Postgres.
> It would be nice if I could use 1 sample per column,(because updating
> individual columns/samples is clear to me) but postgres doesn't compress the
> row (which is bad because of high amount of repetitive data.. Easily 10X
> bigger.
if there are lots of equal readings perhaps you can drop them and only
record when the reading changes.
this would probably require one table per source.
--
⚂⚃ 100% natural