Re: rman ? nfs ? slow

From: Mark D Powell <Mark.Powell_at_eds.com>
Date: Thu, 1 Jan 2009 06:58:21 -0800 (PST)
Message-ID: <b770ee2f-835e-446e-88ed-7dd7f94a9546_at_p2g2000prf.googlegroups.com>



On Jan 1, 7:03 am, "Gerard H. Pille" <g..._at_skynet.be> wrote:
> lfree schreef:
>
>
>
>
>
> > I use rman backup product database on OS windows(2003 R2 standard 64
> > edition ) ,I check to find
> > it is slow (vs Linux rhel 4.3 ). My nfs server use SFU for windows.
>
> > windows:
>
> >  => select * from v$version ;
> > BANNER
> > ----------------------------------------------------------------
> > Oracle Database 10g Enterprise Edition Release 10.2.0.4.0 - 64bi
> > PL/SQL Release 10.2.0.4.0 - Production
> > CORE    10.2.0.4.0      Production
> > TNS for 64-bit Windows: Version 10.2.0.4.0 - Production
> > NLSRTL Version 10.2.0.4.0 - Production
>
> > SELECT start_time, end_time, status, input_type, elapsed_seconds,
> >        compression_ratio, TRIM (input_bytes_display) inputb,
> >        TRIM (output_bytes_display) outputb,
> >        TRIM (input_bytes_per_sec_display) input,
> >        TRIM (output_bytes_per_sec_display) output,
> >        TRIM (time_taken_display) TIME
> >   FROM v$rman_backup_job_details
> >  WHERE start_time > TRUNC (SYSDATE) - 1;
>
> > START_TIME          END_TIME            STATUS           INPUT_TYPE
> > ELAPSED_SECONDS COMPRESSION_RATIO INPUTB   OUTPUTB  INPUT    OUTPUT   TIME
> > ------------------- ------------------- ---------------- ---------- --------
> > ------- ----------------- -------- -------- -------- -------- ----------
> > 2008-12-31 20:03:06 2008-12-31 20:19:37 COMPLETED        ARCHIVELOG
> > 991           2.59272 1.02G    402.88M  1.05M    416.29K  00:16:31
>
> > output_bytes_per_sec_display = 416.29K.
>
> > linux :
>
> > BANNER
> > ----------------------------------------------------------------
> > Oracle Database 10g Enterprise Edition Release 10.2.0.3.0 - 64bi
> > PL/SQL Release 10.2.0.3.0 - Production
> > CORE    10.2.0.3.0      Production
> > TNS for Linux: Version 10.2.0.3.0 - Production
> > NLSRTL Version 10.2.0.3.0 - Production
>
> > START_TIME          END_TIME            STATUS           INPUT_TYPE
> > ELAPSED_SECONDS COMPRESSION_RATIO INPUTB   OUTPUTB  INPUT    OUTPUT   TIME
> > ------------------- ------------------- ---------------- ---------- --------
> > ------- ----------------- -------- -------- -------- -------- ----------
> > 2009-01-01 02:10:17 2009-01-01 02:17:51 COMPLETED        ARCHIVELOG
> > 454           2.72805 2.91G    1.07G    6.56M    2.40M    00:07:34
> > 2008-12-31 02:10:17 2008-12-31 02:16:25 COMPLETED        ARCHIVELOG
> > 368           2.59673 2.25G    888.76M  6.27M    2.42M    00:06:08
>
> > output_bytes_per_sec_display = 2.4M
>
> > in windows, I use to copy command , archive filesiez = 42M ,it takes only
> > 2-3s¡£
> > copy arch... \\192.168.25.100\backup
>
> > why?
>
> > thank advanced.
>
> You are taking a backup on system A (Linux or Whinedoze) while the data is in fact on system B
> (your NFS)?   Not such a good idea.  In which case your problem could be network related.
>
> I don't find it strange that Linux could beat Whinedoze on network throughput.
>
> Gerard- Hide quoted text -
>
> - Show quoted text -

On various versions of UNIX/Linux there are Oracle recommended mount options for use with NFS file systems. (11g comes with an NFS driver built into the Oracle Kernal). Check on your NFS configuration. It could just be that the NFS device is not as fast on Windows as in a UNIX environment or there may be some configuration changes you need to make.

Consider using the parallel option in rman.

HTH -- Mark D Powell -- Received on Thu Jan 01 2009 - 08:58:21 CST

Original text of this message