We have done much work with our file system, volume manager, and disk
array to assure that the best possible IO is 512k. If we use our load
generator and push 512k of data per IO, iostat looks good and it makes
sense. Oracle as well has large io on the same file system. But the
moment we set "filesystemio_options=SETALL" (within oracl) all the io
via iostat goes to 8k and performance is terrible. it is almost like
oracle ignores it multiblock read count parameter.
I'm not a dba but am working closly with them to attempt to resolve
this. Here are some of the solaris parms.
# uname -a
SunOS qacubsdb4 5.10 Generic_137111-07 sun4v sparc
SUNW,SPARC-Enterprise-T5220
# cat /etc/system
* 10/10/2008: (YL) Added for Oracle
set noexec_user_stack=1
* Begin MDD root info (do not edit)
rootdev:/pseudo/m...@0:0,10,blk
* End MDD root info (do not edit)
set maxphys=0x800000
# isainfo -kv
64-bit sparcv9 kernel modules
# prtconf -v | grep Mem
Memory size: 32640 Megabytes
# mkfs -m /dev/md/rdsk/d51
mkfs -F ufs -o
nsect=128,ntrack=48,bsize=8192,fragsize=8192,cgsize=143,free=1,rps=1,nbp
i=997778,opt=t,apc=0,gap=0,nrpos=1,maxcontig=64,mtb=y /dev/md/rdsk/d51
2330840960
# metastat d51
d51: Concat/Stripe
Size: 2330840960 blocks (1.1 TB)
Stripe 0: (interlace: 512 blocks)
Device Start Block
Dbase Reloc
/dev/dsk/c4t600A0B800038AFE10000029448ECB3C7d0s6 0
No Yes
/dev/dsk/c4t600A0B800038AE0B000002A848ECB504d0s6 0
No Yes
Device Relocation Information:
Device Reloc Device ID
/dev/dsk/c4t600A0B800038AFE10000029448ECB3C7d0 Yes
id1,s...@n600a0b800038afe10000029448ecb3c7
/dev/dsk/c4t600A0B800038AE0B000002A848ECB504d0 Yes
id1,s...@n600a0b800038ae0b000002a848ecb504
# cat /etc/vfstab
fd - /dev/fd fd - no -
/proc - /proc proc - no -
/dev/md/dsk/d20 - - swap - no -
/dev/md/dsk/d10 /dev/md/rdsk/d10 / ufs 1 no
logging
/dev/md/dsk/d51 /dev/md/rdsk/d51 /oracle/data ufs 2
yes logging,forcedirectio
/dev/dsk/c4t600A0B800038AE0B0000029E48ECB44Ed0s6
/dev/rdsk/c4t600A0B800038AE0B0000029E48ECB44Ed0s6 /oracle/apps
ufs 2 yes logging
/devices - /devices devfs - no -
ctfs - /system/contract ctfs - no -
objfs - /system/object objfs - no -
swap - /tmp tmpfs - yes -
cubsdev:/export/share - /export/share nfs - yes
ro,nosuid,bg,soft,retry=500
cubsdev:/export/cdrom - /export/cdrom nfs - yes
ro,nosuid,bg,soft,retry=500
cubsdev:/export/cdrom2 - /export/cdrom2 nfs - yes
ro,nosuid,bg,soft,retry=500
cubsdev:/export/cdrom3 - /export/cdrom3 nfs - yes
ro,nosuid,bg,soft,retry=500
cubsdev:/export/cdrom4 - /export/cdrom4 nfs - yes
ro,nosuid,bg,soft,retry=500
cubsrep:/orashare - /orashare nfs - yes
ro,nosuid,bg,soft,retry=500
samqfs1:/qfs1/cubs - /oracle/backups nfs - yes
rw,nosuid,bg,soft,retry=500
--
mike cannon
[email protected] <mailto:[email protected]>
864.650.2577 (cell)
864.656.3809 (office)
computing & information technology
340 computer court
anderson, sc 29625
_______________________________________________
storage-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/storage-discuss
