johnny@edvvie.at (Johann Schweigl) (09/15/89)
I know, in Version 6 everything will be easier, better and faster.
It's correct, we have it (beta).
But we also have a large V5 application in need to be tuned. The database isn't
bad, far from that, but with a given amount of time (and money) to develop the
application most SQL statements deliver the data they should, but not at the
speed they could.
As several hundred of them are hidden in about 70 forms and lots reports,
identifing the slow ones would be too time consuming.
What I would like to have is a means to intercept statements on their way
to the RDBMS kernel, auditing the SQL text and its overall execution time when
the first tuple is returned.
Having this statistics, we could focus the optimization to the most frequently
used or most time consuming statements.
When the TWO_TASK environment variable is set to pipe driver usage, the
application program passes it's data from and to the ORACLE shadow process
through a pipe. If one could manage it to smuggle a dummy process between
them, passing data up and down (like a streams module) but writing selected
pieces to a log file, then, I think, it could be done, independent of the
application (SQL*FORMS, SQL*RPT, SQL*CALC, PRO*C, whatever).
The main problem is to aquire some knowledge about the protocol between
application and shadow process.
Any ideas? Alternate ways? Everything would help.
Thanxalot.
--
------------------------------------------------------------------
EDV Ges.m.b.H Vienna Johann Schweigl
Hofmuehlgasse 3 - 5 USENET: johnny@edvvie.at
A-1060 Vienna, Austria Tel: (0043) (222) 59907 257 (8-19 CET)