Re: setting of audit_trail initialization parameter
From: ~Jeff~ <jifjif_at_gmail.com>
Date: Wed, 22 May 2013 10:01:06 +1200
Message-ID: <CANz7uMaf-b0+R6VDDKLLnrQyW0qRLELaXmsONNi-SaavWz8Bqg_at_mail.gmail.com>
we use DB option here, with a few tweaks:
- for the bigger/most active databases, there is also a AUD$HIST table that
a/ is in its own tablespace (*not* SYSTEM, sheesh!) b/ is compressed c/ gets loaded via a daily append job.
- for the 11g db's, AUD$ is also reconfigured to use an AUDIT tablespace.
Irreversible space blowouts happen to SYSTEM otherwise, when un-movable catalog extents get written at the end of the datafile. Its fairly common to see auditing activity as a top consumer in some databases, and I can imagine it'd be a headache in a larger OLTP environment. Row lock waits in particular.
Date: Wed, 22 May 2013 10:01:06 +1200
Message-ID: <CANz7uMaf-b0+R6VDDKLLnrQyW0qRLELaXmsONNi-SaavWz8Bqg_at_mail.gmail.com>
we use DB option here, with a few tweaks:
- for the bigger/most active databases, there is also a AUD$HIST table that
a/ is in its own tablespace (*not* SYSTEM, sheesh!) b/ is compressed c/ gets loaded via a daily append job.
- for the 11g db's, AUD$ is also reconfigured to use an AUDIT tablespace.
Irreversible space blowouts happen to SYSTEM otherwise, when un-movable catalog extents get written at the end of the datafile. Its fairly common to see auditing activity as a top consumer in some databases, and I can imagine it'd be a headache in a larger OLTP environment. Row lock waits in particular.
I'd imagine OS or XML options might be better, particularly if analysing/reporting on them is not a DBA task :]
cheers
Jeff Wong
-- http://www.freelists.org/webpage/oracle-lReceived on Wed May 22 2013 - 00:01:06 CEST