Re: IO statistics of the ARCH process
From: Lei Zeng <leizeng2003_at_yahoo.com>
Date: Mon, 19 Mar 2012 16:16:29 -0700 (PDT)
Message-ID: <1332198989.25071.YahooMailNeo_at_web114706.mail.gq1.yahoo.com>
Oracle 11g offers some stats data to calculate IOPS and MBPS for ARCH process. However, I am not sure how accurate those database stas could be. As commened in metalink note [1275596.1], " Datafile writes are not as easy to measure as reads" and I assume what ARCH process does is purely 'writes'
Most of the time the ARCH destination is independent volume or mount point, I perfer iostats straight from either OS level or Storage level (or both). It is just a question where to measure at different IO stack layers: Storage <--> OS <---> Database
V$IOSTAT_FUNCTION , FUNCTION_NAME = 'ARCH'
AWR report section
IOStat by Filetype summary
IOStat by Function/Filetype summary
AWR tables in DBA_HIST_IOSTAT_FILETYPE, DBA_HIST_IOSTAT_DETAIL DBspeed sample
Snap_id End_Snap_time Inst# Filetype_name Read_IOPS Read_MBPS Avg_small_read_time_ms Avg_large_read_time_ms Write_IOPS Write_MBPS
Lei
>From: Tanel Poder <tanel_at_tanelpoder.com>
>To: Amir.Hameed_at_xerox.com
>Cc: oracle-l_at_freelists.org
>Sent: Friday, March 16, 2012 4:01 PM
>Subject: Re: IO statistics of the ARCH process
>>Snapper can show it as arch processes have sessions (and populate
>V$SESSION) like most other background processes...
>
>*Tanel Poder*
>Enkitec Europe
>+372 56 956 181
>http://www.enkitec.com/
>Expert Oracle Exadata book:
>http://www.apress.com/9781430233923
>
>On Fri, Mar 16, 2012 at 7:20 PM, Hameed, Amir <Amir.Hameed_at_xerox.com> wrote:
>
> Folks,
> Is there a statistic that can be used to estimate how much IO/sec the
> ARCH process is doing when archiving redo log files? I was planning to
> use "Log archive I/O" event but the description of this event states
> that:
>
> "When the archiving process exhausts its I/O buffers because all of them
> are being used for on-going I/O's, the wait for an available I/O buffer
> is captured in this system wait event."
>
>
>
> Thank you
>
> Amir
>
>
Date: Mon, 19 Mar 2012 16:16:29 -0700 (PDT)
Message-ID: <1332198989.25071.YahooMailNeo_at_web114706.mail.gq1.yahoo.com>
Oracle 11g offers some stats data to calculate IOPS and MBPS for ARCH process. However, I am not sure how accurate those database stas could be. As commened in metalink note [1275596.1], " Datafile writes are not as easy to measure as reads" and I assume what ARCH process does is purely 'writes'
Most of the time the ARCH destination is independent volume or mount point, I perfer iostats straight from either OS level or Storage level (or both). It is just a question where to measure at different IO stack layers: Storage <--> OS <---> Database
V$IOSTAT_FUNCTION , FUNCTION_NAME = 'ARCH'
AWR report section
IOStat by Filetype summary
IOStat by Function/Filetype summary
AWR tables in DBA_HIST_IOSTAT_FILETYPE, DBA_HIST_IOSTAT_DETAIL DBspeed sample
Snap_id End_Snap_time Inst# Filetype_name Read_IOPS Read_MBPS Avg_small_read_time_ms Avg_large_read_time_ms Write_IOPS Write_MBPS
------- -------------------- ---- ------------- -------- -------- ---------------------- ---------------------- ---------- ---------- 4-5 2012-03-19 01:00:27 1 Archive Log 0 0 0 0 0.73 0.735-6 2012-03-19 02:00:53 1 Archive Log 0 0 0 0 1.11 1.1 6-7 2012-03-19 03:00:13 1 Archive Log 0 0 0 0 0.72 0.72
Lei
>From: Tanel Poder <tanel_at_tanelpoder.com>
>To: Amir.Hameed_at_xerox.com
>Cc: oracle-l_at_freelists.org
>Sent: Friday, March 16, 2012 4:01 PM
>Subject: Re: IO statistics of the ARCH process
>>Snapper can show it as arch processes have sessions (and populate
>V$SESSION) like most other background processes...
>
>*Tanel Poder*
>Enkitec Europe
>+372 56 956 181
>http://www.enkitec.com/
>Expert Oracle Exadata book:
>http://www.apress.com/9781430233923
>
>On Fri, Mar 16, 2012 at 7:20 PM, Hameed, Amir <Amir.Hameed_at_xerox.com> wrote:
>
> Folks,
> Is there a statistic that can be used to estimate how much IO/sec the
> ARCH process is doing when archiving redo log files? I was planning to
> use "Log archive I/O" event but the description of this event states
> that:
>
> "When the archiving process exhausts its I/O buffers because all of them
> are being used for on-going I/O's, the wait for an available I/O buffer
> is captured in this system wait event."
>
>
>
> Thank you
>
> Amir
>
>
DBspeed http://dbspeed.com/case_study.html
-- http://www.freelists.org/webpage/oracle-lReceived on Mon Mar 19 2012 - 18:16:29 CDT