I can't drop the table. I have to add as many columns as posible and when I exceed the limit I have to create another table.
I've tried normalizing but then the join's cost is too big. I always need to use all columns, so getting a all information into a single row it's the most efficient solution. On Wed, Jul 13, 2011 at 1:51 PM, Scott Marlowe <scott.marl...@gmail.com>wrote: > On Wed, Jul 13, 2011 at 9:45 AM, Miguel Angel Conte <diaf...@gmail.com> > wrote: > > Hi, > > Thanks for your interest. This app load scv files which change every day > > (sometimes the columns too). The sizes of these files are in avg 15MB. > So, > > We load something like 100MB each day. We tried to find a better solution > > but we couldn't, becouse one of the our requirement is not to use a lot > of > > space. Also, the app is used to consult these information, and for our > > particular type of select's queries, we get the best performance if the > > information is all into a same row. > > This definitely sounds like a job for something like hstore. > > Also, can you drop and recreate the table every so often? That would > certainly help. >