On Friday 18 December 2009 11:00:33 am sabrina miller wrote: > Hi everybody, > My requirements was: > + Made a table charge to be partitioned by carrier and month > + summarize by charges > + summarize by users, > + each summarization must be by month and several others columns. > > > > Doesn't sound like too much? As I say, im new and I didn't found any > better. But an insert takes around 135ms in the worst case (create tables > and insert rows) and about 85 ms in best case (only updates). There are > something better? If I am following this it means there is an average of 50ms extra overhead to do an INSERT on charges.charges then an UPDATE correct? If so you have to consider that an INSERT is actually doing quite a lot besides creating a new row in charges.charges. There is a time cost to querying the database for existence of objects , making decisions based on the result, creating new database objects and the populating those objects. The issue then becomes where you want to pay it? So the something better question then becomes where is the best place to incur that cost. If the 135ms worst case works and does not impede your process then it may be the best solution. Unfortunately there is not enough information to give a definitive answer. > > Thanks in advance, Sabrina -- Adrian Klaver aklaver@xxxxxxxxxxx -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
