Thread: Savepoints in ODBC driver
Hello, while checking some performance bottlenecks I realized that the ODBC driver is creating/deleting a savepoint for *every* command send to the server. Grep for SC_start_rbpoint and SC_started_rbpoint to see what I mean. This does require 3 roundtrips to the server (instead of one) and some additional processing time for the backend to create/delete the savepoint every time. What's the rational behind this? And can this be turned off somewhere? Rainer
Hello, anybody? Sunday, August 9, 2009, 10:38:48 PM, I wrote: RB> Hello, RB> while checking some performance bottlenecks I realized that the ODBC RB> driver is creating/deleting a savepoint for *every* command send to RB> the server. RB> Grep for SC_start_rbpoint and SC_started_rbpoint to see what I mean. RB> This does require 3 roundtrips to the server (instead of one) and some RB> additional processing time for the backend to create/delete the RB> savepoint every time. RB> What's the rational behind this? And can this be turned off somewhere? RB> Rainer
On Tue, Aug 18, 2009 at 12:45 PM, Rainer Bauer<usenet@munnin.com> wrote: > RB> while checking some performance bottlenecks I realized that the ODBC > RB> driver is creating/deleting a savepoint for *every* command send to > RB> the server. I can only guess. But I would expect that your "Level of Rollback on Error" could be causing this. You might try a different setting. -- Regards, Richard Broersma Jr. Visit the Los Angeles PostgreSQL Users Group (LAPUG) http://pugs.postgresql.org/lapug
Hello Richard, Tuesday, August 18, 2009, 11:26:49 PM, you wrote: RB> On Tue, Aug 18, 2009 at 12:45 PM, Rainer Bauer<usenet@munnin.com> wrote: >> RB> while checking some performance bottlenecks I realized that the ODBC >> RB> driver is creating/deleting a savepoint for *every* command send to >> RB> the server. RB> I can only guess. But I would expect that your "Level of Rollback on RB> Error" could be causing this. You might try a different setting. Bingo! Thanks Richard. Rainer