detecting table scans

Hello,

In a huge complex Sybase application written in ESQL/C and JDBC with hundreds
of tables and some hundred concurrent connects to ASE 12.5.4 we're facing situations
in which a simple FETCH from a small table, which normally takes 5-6 ms, takes
from time to time 10-20 ms, or INSERT lasts even 3 secs; it is unclear what is
causing this. One of our hypothesis is that the application in some areas, sometimes,
is falling into full table scans, and the above mentioned long INSERT is victim of such
an event of other processes running at the same time.

My question is: Is there some kind of logging I could switch on (for example by a trace
flag) which will let ASE log all such table scans?

TIA

matthias
--
http://www.unixarea.de/

On Nov 21, 5:26 am, Matthias Apitz wrote:
> Hello,
>
> In a huge complex Sybase application written in ESQL/C and JDBC with hundreds
> of tables and some hundred concurrent connects to ASE 12.5.4 we're facing situations
> in which a simple FETCH from a small table, which normally takes 5-6 ms, takes
> from time to time 10-20 ms, or INSERT lasts even 3 secs; it is unclear what is
> causing this. One of our hypothesis is that the application in some areas, sometimes,
> is falling into full table scans, and the above mentioned long INSERT is victim of such
> an event of other processes running at the same time.
>
> My question is: Is there some kind of logging I could switch on (for example by a trace
> flag) which will let ASE log all such table scans?
>
> TIA
>
>         matthias
> --http://www.unixarea.de/

Perhaps the MDA monitoring table monSysPlanText in conjuntion with
monSysSQLText?

-bret