Re: ADDM advices Oracle's application logic
From: Frank van Bortel <frank.van.bortel_at_gmail.com>
Date: Fri, 18 Jan 2008 15:53:31 +0100
Message-ID: <95d02$4790bd6b$524b5c40$3179@cache6.tilbu1.nb.home.nl>
>> Helma wrote:
>>> After i recreated the filesystem and put the databack on it, i ran the
>>> command for an extra datafile with the 10046 trace, and it took 8
>>> minutes, which seems acceptable for 16Gb. ( how fast are yours?)
>> Similar effect here. CentOS 5.1 plus LVM, increased
>> extent from (std) 32 to 64M. Two Samsung HD321KJ SATA disks.
>> 4GB memory, Intel E6600, running 2.52 GHz, Oracle 10.2.0.3, 32 bit.
>> No tracing - creating 4 tablespaces @16GB each:
>>
>> Tablespace created.
>> Elapsed: 00:05:16.44
>>
>> Tablespace created.
>> Elapsed: 00:05:40.95
>>
>> Tablespace created.
>> Elapsed: 00:05:48.19
>>
>> Tablespace created.
>> Elapsed: 00:06:19.63
>>
>> 1 minute increase for the 4th ts.
>> Dropping (including contents and datafiles) seems to go
>> faster with less tablespaces to do:
>> Tablespace dropped.
>> Elapsed: 00:00:32.26
>>
>> Tablespace dropped.
>> Elapsed: 00:00:29.47
>>
>> Tablespace dropped.
>> Elapsed: 00:00:26.20
>>
>> Tablespace dropped.
>> Elapsed: 00:00:19.77
>>
>> I have no idea how to start explaining this.
>> --
>>
>> Regards,
>> Frank van Bortel
>>
>> Top-posting in UseNet newsgroups is one way to shut me up
Date: Fri, 18 Jan 2008 15:53:31 +0100
Message-ID: <95d02$4790bd6b$524b5c40$3179@cache6.tilbu1.nb.home.nl>
Helma wrote:
> On 18 jan, 11:01, Frank van Bortel <frank.van.bor..._at_gmail.com> wrote:
>> Helma wrote:
>>> After i recreated the filesystem and put the databack on it, i ran the
>>> command for an extra datafile with the 10046 trace, and it took 8
>>> minutes, which seems acceptable for 16Gb. ( how fast are yours?)
>> Similar effect here. CentOS 5.1 plus LVM, increased
>> extent from (std) 32 to 64M. Two Samsung HD321KJ SATA disks.
>> 4GB memory, Intel E6600, running 2.52 GHz, Oracle 10.2.0.3, 32 bit.
>> No tracing - creating 4 tablespaces @16GB each:
>>
>> Tablespace created.
>> Elapsed: 00:05:16.44
>>
>> Tablespace created.
>> Elapsed: 00:05:40.95
>>
>> Tablespace created.
>> Elapsed: 00:05:48.19
>>
>> Tablespace created.
>> Elapsed: 00:06:19.63
>>
>> 1 minute increase for the 4th ts.
>> Dropping (including contents and datafiles) seems to go
>> faster with less tablespaces to do:
>> Tablespace dropped.
>> Elapsed: 00:00:32.26
>>
>> Tablespace dropped.
>> Elapsed: 00:00:29.47
>>
>> Tablespace dropped.
>> Elapsed: 00:00:26.20
>>
>> Tablespace dropped.
>> Elapsed: 00:00:19.77
>>
>> I have no idea how to start explaining this.
>> --
>>
>> Regards,
>> Frank van Bortel
>>
>> Top-posting in UseNet newsgroups is one way to shut me up
> > hi Frank, > > We use 432320-001 serial scsi 10k rpm disk HP/Compaq. Specs: > > http://www.yobitech.com/ProductDetails.asp?ProductCode=432320%2D001&Show=TechSpecs > > $320 > > # Performance > # Drive Transfer Rate 300 MBps (external) > # Seek Time 4 ms (average) / 8.1 ms (max) > # Spindle Speed 10000 rpm > # Track-to-Track Seek Time 0.2 ms > > this seems to be better specs than yours: > > http://www.samsung.com/nl/products/hdd/spinpointtseries/hd321kj.asp > pinpoint T Serie > HD321KJ > Rotational speed 7200 rpm > Track to track 0.8 ms > Average 8.9 ms > Full stroke 18 ms > Average latency 4.17 ms > > Yet you have the better performance. I know very little of hardware, > does anyone have an idea? > > H.
SATA... Although I think my drives are capped at 1.5Gbps, SATA can handle up to 3 Gbps - the rest is of no importance for this continuous write action - just (sustained) write.
And... I purchased those disks for a little over 100 Eur. almost a year ago. Current price levels are about 85 Eur for 500GiB.
Beats those branded scsi's any time, price/performance wise. I'll see if I can get some (write) benchmarks on this system.
-- Regards, Frank van Bortel Top-posting in UseNet newsgroups is one way to shut me upReceived on Fri Jan 18 2008 - 08:53:31 CST