<manish.nevrekar_at_gmail.com> wrote in message
news:1107294990.249825.185300_at_c13g2000cwb.googlegroups.com...
>I have a large table approx 50 million rows. I have partitioned the
> table by date. I have a bitmapped index on one of the columns and the
> tabel also has a primary key with partitioned column forming the
> leading column of the primary key index. It is in a Datawarehouse, we
> have only inserts and very few updates. The insertion of data ( approx
> 150K rows) takes a lot of time and run ut of space for the rollback
> segments(RBS tablespace ( I have tried allocating upto 4gb for the RBS
> tablespace). I can only finish the loading by introducing commit after
> every 20000 rows( I know it is a bad idea!!!!)
>
> I am using 8.1.7.4 on HP-Tru64. Any ideas/suggestions on what might
> be causing so much rollback?? and what i can do to correct it??
> TIA
> Manish
>
How wide are your rows? Do they contain *LOBS? Are any other transactions
running? Do you have triggers that are doing work on the inserts?
Personally, I don't think committing on every x rows is a bad idea, to the
contrary, I think it's a good idea for the reason you just stated. If you
are worried about rolling out a bad ETL, make note of the last good record
and remove the bad rows.
--
Tom Dyess
OraclePower.com
Received on Wed Feb 02 2005 - 07:32:15 CST
