Flashback Table Times

[dnunknown]

Hi Tim,

11.2.0.1.0

I was wondering what parameters should be looked into for increasing performance of Flashback, we are flashing back a series of tables

I am testing on a table that is 2.3M rows, its size is 346MB, it is taking 6 minutes to delete all the data w/ row movement enabled and 5 minutes to flash it back

I have ensured there is enough room in the FRA and the db_recover_file_dest_size, also increased the number of redo log groups. There are no constraints or indexes on the table

A couple of blogs are saying log_buffer should be increased, any thoughts?

Of course we are only going to be as fast as the disk presented to the FRA

Thanks,
Steve

[Tim...]

Hi.

I guess the obvious question is, "What is using all the time?".

Before you know what to do to fix anything, you need to know the root cause of the performance issue. What are the predominant wait states associated with the flashback operation? If you trace it, you will know where the time is being spent and that will give you an idea of what you can do to improve the performance.

Cheers

Tim...