Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Mailing Lists -> Oracle-L -> Re: my initial storage division for SAN

Re: my initial storage division for SAN

From: rahul sharma <rahul_at_infotech.co.id>
Date: Tue, 29 Apr 2003 01:41:36 -0800
Message-ID: <F001.0058BC7C.20030429014136@fatcity.com>


Arun, dosnt raid0+1 (or 1+0) give better performance compared to RAID5 ? at least that's what i have heard.. and read, if you have money go for stripe and mirror rather than
a RAID5 volume,. NOw my problem is that i would require four such volumes !! data index for isntance A
and data index for instance B !! which is going to take a lot of space !! as the minimum capacity drive in EMC
is 36GB !

> Hi Rahul.
>
> I m not sure why do you have to do Raid(0 +1) for index and data
> tablespaces, you might as well do Raid 5.
> If you either do Raid 0 or Raid 1 then its worth comparing between those
two
> choices i.e. with Raid 5.
>
> Other fact is that if you have Raid 5, it doesnt matter where you place
your
> datafiles or logfiles, from the "system admin perspective" its writing to
> all disks(refer to concepts of Raid 5). Only it matters when you strip
using
> volume manager. Are you using veritas volumne manager?
>
> -Arun.
>
> ----- Original Message -----
> To: "Multiple recipients of list ORACLE-L" <ORACLE-L_at_fatcity.com>
> Sent: Monday, April 28, 2003 2:11 AM
>
>
> chao, my reason to let the archived logs from both the instances on one
> volume is that, instance B is not very
> busy, and i do not expect any kind of contection between arch processes of
A
> & B using the same volume
>
> SAME is another concept which i have never tried.. i will go thru it
also..
> thanks for reminding ;-))
>
>
> ----- Original Message -----
> To: "Multiple recipients of list ORACLE-L" <ORACLE-L_at_fatcity.com>
> Sent: Monday, April 28, 2003 1:06 PM
>
>
> > rahul sharma,
> > hi, one question, 300GB to place archivelogs from a&b, they should be
> seperated , right? unless you use cluster filesystem, you cannot let the
two
> node mount the same volume, right?
> > and maybe SAME rule works for you, consider it.And whether your system
is
> write intensive, you can check it like:
> >
> > 13:08:50 perfstat_at_BIDDB.EACHNET.COM> @readwrite
> >
> > Total buffer
> > -------------
> > 1.6515E+10
> >
> > Elapsed: 00:00:00.48
> > old 2: :totalcost := &totalcost;
> > new 2: :totalcost := 1.6515E+10;
> >
> > PL/SQL procedure successfully completed.
> >
> > Elapsed: 00:00:00.01
> >
> > total execution
> > ---------------
> > 1218693220
> >
> > Elapsed: 00:00:00.29
> > old 2: :totalexec :=&totalexec;
> > new 2: :totalexec :=1218693220;
> >
> > PL/SQL procedure successfully completed.
> >
> > Elapsed: 00:00:00.02
> >
> > SQL_TYPE Total buffer total execution average Cost 成本比率
> 执行比率
>
> ---------- ------------- --------------- ------------ ---------- ---------
> -
> > BEGIN 2848457875 232752041 12.238165 17.24
> 19.09
> > DECLA 218094451 6762742 32.2494117 1.32
> .55
> > DELET 47111763 1617433 29.1274897 .28
> .13
> > INSER 320604425 41939893 7.64437871 1.94
> 3.44
> > SELEC 1.2343E+10 880118415 14.0239731 74.73
> 72.21
> > UPDAT 737797721 55448104 13.3060947 4.46
> 4.54
> >
> > column totalcost heading "Total buffer " new_value totalcost
> > select sum(buffer_gets) totalcost from v$sql;
> > variable totalcost number;
> >
> > begin
> > :totalcost := &totalcost;
> > end;
> > /
> >
> > col totalexec heading "total execution" new_value totalexec
> > select sum(executions) totalexec from v$sql;
> > variable totalexec number;
> > begin
> > :totalexec :=&totalexec;
> > end;
> > /
> >
> > select upper(substr(sql_text,1,5)) SQL_TYPE,sum(buffer_gets) TotalCost
> ,sum(executions) TotalExec
> > ,sum(buffer_gets)/sum(executions) "average Cost",
> > trunc(Sum(buffer_gets)/(:totalcost),4)*100 "成本比率",
> > Trunc(sum(executions)/(:totalexec),4)*100 "执行比率"
> > from v$sql
> > group by upper(substr(sql_text,1,5))
> > having sum(buffer_gets)>100000;
> >
> >
> >
> >
> > >we are in a process of migratin our instances to SAN, i have been given
> the
> > >task of
> > >dividing the SAN storage :-((
> > >
> > >here are my initial setup, we have been given a 1TB box ... please let
me
> > >know if i am missing something
> > >
> > >(two instances A & B)
> > > - 36GB x 2 for redo logs 3 groups of two members each 500M size,
> > >multiplexed instance A
> > > - 36GB x 2 for redo logs 3 groups of two members each 500M size,
> > >multiplexed instance B
> > >- 300GB to hold archive logs from A & B , mirrored
> > >- 36GB drive for TEMP A & B
> > >
> > >the rest of the storage will be divided into DATA and INDEX for A & B ,
> my
> > >question is.. should we make
> > >RAID5 for data and index ?? or individual drives ? we know the IO
> patterns
> > >for all the tablespaces, ideally the write intensive tbs should be
> RAID(1+0)
> > >, but we might not have enough space if we do that. !! (have i answerd
> all
> > >my questions ??? ;-))
> > >
> > >i would appreciate any comments from you guys who have DB's running on
> SAN.
> > >
> > >TIA
> > >
> > >
> > >
>
>
> --
> Please see the official ORACLE-L FAQ: http://www.orafaq.net
> --
> Author: rahul sharma
> INET: rahul_at_infotech.co.id
>
> Fat City Network Services -- 858-538-5051 http://www.fatcity.com
> San Diego, California -- Mailing list and web hosting services
> ---------------------------------------------------------------------
> To REMOVE yourself from this mailing list, send an E-Mail message
> to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
> the message BODY, include a line containing: UNSUB ORACLE-L
> (or the name of mailing list you want to be removed from). You may
> also send the HELP command for other information (like subscribing).
>
>
>
>
> --
> Please see the official ORACLE-L FAQ: http://www.orafaq.net
> --
> Author: Arun Annamalai
> INET: oracle_at_shakaboom.com
>
> Fat City Network Services -- 858-538-5051 http://www.fatcity.com
> San Diego, California -- Mailing list and web hosting services
> ---------------------------------------------------------------------
> To REMOVE yourself from this mailing list, send an E-Mail message
> to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
> the message BODY, include a line containing: UNSUB ORACLE-L
> (or the name of mailing list you want to be removed from). You may
> also send the HELP command for other information (like subscribing).
>
>

-- 
Please see the official ORACLE-L FAQ: http://www.orafaq.net
-- 
Author: rahul sharma
  INET: rahul_at_infotech.co.id

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).
Received on Tue Apr 29 2003 - 04:41:36 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US