Re: Redo per transaction inconsistency when running SLOB

As far as I know "redo size" is the sum of the redo byte written to the log buffer by the invidivual sessions.

The log writer reports redo blocks written, and the difference between redo size and redo blocks written (typically 512 byte blocks - though 4K is now possible) is reported as the "redo wastage";

Then, I think, the "redo block written" is "logical" blocks, not physical blocks, i.e. I don't think it would be multiplied by 3 if you had 3 files per log group - consequently I don't THINK writes to remote log files would show up in these older statistics.

Regards
Jonathan Lewis

On Thu, Apr 23, 2020 at 5:00 PM Tim Gorman <tim.evdbt_at_gmail.com> wrote:

> Mightn't "redo size" be affected by the number of log archive destinations?
>
> If the standby is MAX PROTECTION or MAX AVAILABILITY , then LGWR
> process(es) have to write out to the standby along with writes down to the
> online redo log files.
>
> I can't imagine any process other than LGWR to calculate and store the
> "redo size" statistic, so my question is whether LGWR tallies bytes read
> from the Log Buffer, or whether it tallies bytes written to the
> destinations (i.e. online redo plus MAX PROT/AVAIL standby dests)?
>
>
>

--
http://www.freelists.org/webpage/oracle-l