> -----Original Message----- > From: Yan Cheng Cheok [mailto:yccheok@xxxxxxxxx] > Sent: Tuesday, January 05, 2010 10:30 PM > To: Craig Ringer > Cc: Dann Corbit; pgsql-general@xxxxxxxxxxxxxx > Subject: Re: PostgreSQL Write Performance > > Thanks for the information. I perform benchmarking on a very simple > table, on local database. (1 table, 2 fields with 1 is bigserial, > another is text) > > ==================================================================== > INSERT INTO measurement_type(measurement_type_name) VALUES ('Hello') > ==================================================================== > > I turn synchronous_commit to off. > > To write a single row(local database), the time taken is in between > 0.1ms and 0.5ms > > I try to compare this with flat text file. > > To write a single row(file), the time taken is in between 0.005ms and > 0.05ms > > The different is big. Is this the expected result? Are you guys also > getting the similar result? Ten to one is not surprising. You will save quite a bit by doing a prepare and a large number of inserts in a single transaction. > I know there shall be some overhead to write to database compared to > flat text file. (network communication, interpretation of SQL > statement...) However, Is there any way to further improve so that > PostgreSQL write performance is near to file? > > If not, I need to plan another strategy, to migrate my flat text file > system, into PostgreSQL system smoothly. Did you try the copy command? It will be very much faster. -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
