Missing record in archive log

Question

hello all,
to the point, i am trying to mirroring 2 database using archive log.
so, i do some test by inserting large data (1200 record / s) into 4 table for some interval (like 10 menit).
Each time database create archive log, my daemon in Master will copy the archive log to the Slave, and daemon in the Slave will extract them using dbms_logmnr.
the problem is number of rows in all Slave table is not equal (less than) number of all rows in Master.
even if i try to alter the system log file manuallly, still there is missing rows in slave.
i have try some alternative like,
- increasing db_writer_processes (from 1 to 2 to 4)
- increasing redo log size (from 50 MB to 100 MB)
- increasing redo log group (from 3 to 6)
but still there is a missing record.
where is those missing record go ?
master db: 10.2.0.1.0 + ASM
master os: SunOS z8 5.10 Generic_137138-09 i86pc
slave db: 10.2.0.1.0
slave os: GNU/Linux 2.6.18-8.el5 x86_64 (RHEL)

Answer 1

Thank you for your reply sir.
The problem with CDC (and/or Streams implementation) is they use a cpu resource within Oracle it self. So, if in peak time Oracle in Master has already reaching 80-90%, with addition using Streams they will make it to > 90%. Don't imagine 1000000 / day transaction here, but imagine with 1000/s transaction. And That's what the-one-who-already-use-stream says.
i am not reinvent the wheel here, i am seeking for best alternative. in fact i am looking for answer from "where is those missing record go ?"