Re: RMAN question(s)
- From: "Jared Still" <jkstill@xxxxxxxxx>
- To: Brandon.Allen@xxxxxxxxxxx
- Date: Mon, 17 Sep 2007 18:32:38 -0700
On 9/17/07, Allen, Brandon <Brandon.Allen@xxxxxxxxxxx> wrote:
>
> In case anyone is interested - I did a test of the 10g "as compresssed"
> backupset on a 285GB databases and here are the results:
>
> ...
> So, I can't really make any conclusions on the backup time yet due to
> the large variance in backup times on this environment, but it's clear
> that the space savings was significant (280-36=244 & 244/280=87%
> reduction) and the runtime from this one sample was close to the bottom
> of the normal range so that's promising.
>
The data in v$backup_sync_io and v$backup_async_io may provide better
information after the fact. I too find it difficult to get a system
suitable for testing.
The data in these views is not persistent, so you need to get it after the
backup
completes and save it.
I've pasted a script at the end of this post to get you started with it, if
you
would like to use it.
--
Jared Still
Certifiable Oracle DBA and Part Time Perl Evangelist
--------------------------------------------------
-- rman_io_stats.sql
-- aggregation of compatible stats
-- from the v$backup_async_io and v$backup_sync_io views
--
alter session set nls_date_format = 'mm/dd/yy hh24:mi:ss';
col device_type format a10 head 'DEVICE|TYPE'
col type format a15
col filename format a40 head 'FILENAME' TRUNC
col buffer_size format 99,999,999,999 head 'BUFFER SIZE'
col buffer_count format 9999 head 'BFFR|CNT'
col open_time format a18 head 'OPEN TIME'
col close_time format a18 head 'CLOSE TIME'
col elapsed_time format 999,999 head 'ELAPSED|SECONDS'
col set_stamp format 9999999999 head 'SET STAMP'
col set_count format 99999 head 'SET |COUNT'
col megabytes format 99,999,999.0 head 'MEGABYTES'
col total_megabytes format 99,999,999.0 head 'TOTAL|MEGABYTES'
col effective_bytes_per_second format 999,999,999 head 'EFF BYTES|PER
SECOND'
col io_count format 999,999,999 head 'IO COUNT'
set line 200
break on set_stamp skip 1 on set_count
--spool rio.txt
SELECT
set_stamp
, set_count
, open_time
--, close_time
, device_type
, type
, filename
, buffer_size
, buffer_count
, bytes / 10485760 megabytes
, total_bytes / 10485760 total_megabytes
, effective_bytes_per_second
, io_count
, elapsed_time/100 elapsed_time -- stored in centiseconds
FROM v$backup_async_io -- disk
union
SELECT
set_stamp
, set_count
, open_time
--, close_time
, device_type
, type
, filename
, buffer_size
, buffer_count
, bytes / 10485760 megabytes
, total_bytes / 10485760 total_megabytes
, effective_bytes_per_second
, io_count
, elapsed_time/100 elapsed_time -- stored in centiseconds
FROM v$backup_sync_io -- tape, probably. could be disk
ORDER BY 1,2,3
/
--spool off
--ed rio.txt
Other related posts:
