Hello,
Where i should put data of SAPDB for the best performance and solution in Linux ?
Should i create different partition to exclusively put data SAPDB ?
Or just follow the SAPDB default config in the directory of /opt/sapdb
It depends.
If you only have one disk (mirrored naturally) then it's probably best to use two files (for LOG and DATA) as that would cause less head-movement than putting the data on two different partitions, however, it's not going to be fast.
I've had great success with a 6 disk setup: 2 disks for the filesystem, mirrored
2 disks for the LOG, software mirrored in raw mode 2 disks for the DATA, software mirrored in raw mode
That means that each type of diskacess is on a separate set of spindles and that minimizes the time lost seeking around.
If you can only afford 4 disks put DATA and the filesystem together as the most IO happens to the LOG devices.
Always use the fastest possible disks for the LOG mirror, never use RAID 5 as that kills performance.
Mirroring your data keeps you running even if the box throws a disk, but it also doubles your read performance which is important, write performance is roughly the same as one disk.
Never put two disks of the same mirror on the same controller channel, our latest setup is:
/proc/ide/hda/model: MAXTOR 6L040J2 Disk /dev/hda: 255 heads, 63 sectors, 4866 cylinders
/proc/ide/hdc/model: MAXTOR 6L040J2 Disk /dev/hdc: 16 heads, 63 sectors, 77557 cylinders
/proc/ide/hde/model: WDC WD1200JB-75CRA0 Disk /dev/hde: 16 heads, 63 sectors, 232581 cylinders
/proc/ide/hdf/model: ST340016A Disk /dev/hdf: 16 heads, 63 sectors, 77545 cylinders
/proc/ide/hdg/model: ST340016A Disk /dev/hdg: 255 heads, 63 sectors, 4865 cylinders
/proc/ide/hdh/model: WDC WD1200JB-75CRA0 Disk /dev/hdh: 16 heads, 63 sectors, 232581 cylinders
Kernel parameters, needed because of insane bios'es that think that they need to make the same disk appear with different c,h,s depending on the location, hint: dev=c,h,s
hdc=4866,255,63 hde=232581,16,63 hdh=232581,16,63 hdf=77545,16,63 hdg=77545,16,63
partitions: File system: hda1 full disk hdc1 full disk
Data: hde1 2GB hdh1 2GB
LOG: hdf1 256MB hdg1 256MB
Swap (striped): hde2 1GB hdh2 1GB
backup hde3 100GB hdh3 100GB
mirrors: md0 : hda1 + hdc1 : / md1 : hde1 + hdh1 md2 : hdf1 + hdg1 md3 : hde3 + hdh3 : /backup
raw devices: raw/raw1 : md1 : SAPDB Data space raw/raw2 : md2 : SAPDB Log space
-- Regards Flemming Frandsen - http://dion.swamp.dk PartyTicket.Net co founder & Yet Another Perl Hacker
-- MaxDB Discussion Mailing List For list archives: http://lists.mysql.com/maxdb To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
