Re: dbd::oracle behavior

Ouch Sorry late reply on this one

If you are using an earlier version of DBD::Oracle and Apache::DBI  you 
might be seeing the 'PING' effect.

To keep connections alive in Apache::DBI it uses the $h->ping  which in 
older DBD does a select on a table.

260 and 360 are not very many.  I have seen rates well into the 10000s 
for some other users

version 1.22+ fixed this by doing either an OCIPing if you are using 10+ 
client or another OCI call that does a round trip to the server.

Hope this helps

cheers
John Scoles

Matthew Maglinte wrote:
> I'm a dba supporting a  warehouse database (production environment) 
> accessed via a custom application built on open source components, one of 
> which is dbd::oracle.
>
> Performance monitoring shows two queries continually consuming 
> considerable resources (cpu waits) - the query text resembling for the 
> most part 'column_info' and 'table_info'.  Over any given day, the 
> column_info select is executed +260 times an hour,  'table_info'  executed 
> +360 times an hour.
>
> Is this level of frequency intended?  And are these calls appropriate 
> (expected) for a static data dictionary environment?
>
> Any response that would help my understanding would be appreciated. 
> Thanks.
>
> Matt Maglinte
> DBA - PeMS/LCS
> Department of Transportation
> W:(916)445-0090
> [email protected]
>
>