Re: odbc vs. libpq performance
От | Vilches, Alejandro |
---|---|
Тема | Re: odbc vs. libpq performance |
Дата | |
Msg-id | 9CE034E149417949A58AA9A4FA7E1C5584AF2B48@ORSMSX109.amr.corp.intel.com обсуждение исходный текст |
Ответ на | Re: odbc vs. libpq performance (Heikki Linnakangas <hlinnaka@iki.fi>) |
Список | pgsql-odbc |
(Sorry, didn't reply to the mailing list initially.) Oh, great! Thanks for letting me know! Now, even when setting auto commit to off, performance is still significantly lower (compared to libpq). For example, runningthe sample program I provided earlier on my machine (and setting auto commit to off), I'm a getting ~100 transactionsper second. I did see an improvement compared to before (when I wasn't able to set auto commit to off) as Iwas only getting ~4 transactions per second. However, if I change the program to use libpq, I get ~1400 transactions persecond. I was expecting to see better performance, but the difference shouldn't be too significant, right? What performancedifference should I expect when using ODBC vs libpq? Thanks! Alejandro -----Original Message----- From: Heikki Linnakangas [mailto:hlinnaka@gmail.com] On Behalf Of Heikki Linnakangas Sent: Saturday, August 1, 2015 11:07 AM To: Vilches, Alejandro; 'pgsql-odbc@postgresql.org' Subject: Re: [ODBC] odbc vs. libpq performance On 07/31/2015 02:33 AM, Vilches, Alejandro wrote: > I've continued looking into this and I discovered something: I figured > out how to successfully set the SQL_ATTR_AUTOCOMMIT attribute to off. > If you set it prior to establishing the connection, this setting is > ignored and it defaults to on. If you set it after establishing the > connection, then it does recognize the setting and it works as > specified. > > The ODBC documentation specifies that this attribute can be set either > before or after establishing the connection. If it's not accepting > the setting before establishing the connection, is this an issue in > the PostgreSQL ODBC driver? A-ha! Yes, that's a bug in the driver. In fact, it was fixed back in January: http://www.postgresql.org/message-id/BY1PR0401MB143087AD121D0251C00E5045A85B0@BY1PR0401MB1430.namprd04.prod.outlook.com That fix will be included in the next release. - Heikki
В списке pgsql-odbc по дате отправления: