On Thu, 28 Oct 1999, Stanley, Jeremy wrote:
> If I knew of a SEARCHABLE archive for this list, I'd cite a specific
> example. Suffice it to say that Ingo commented on this about a month
> ago, saying that some of the new changes to the RAID-1 code may have
> caused a worse-than-single-disk degradation in performance and was being
> looked into, with a request for further examples/demonstrative
> benchmarks and test-cases from the peanut gallery...
Here is my results for Raid-1 read performance weirdness:
Raid 1: Chunk size 4k: /dev/md1:
-------Sequential Output-------- ---Sequential Input-- --Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU
raid1ch4 1000 1651 97.1 15473 81.2 6408 54.8 1769 99.4 8317 30.5 136.2 6.4
Non raid partition: /dev/sda4:
-------Sequential Output-------- ---Sequential Input-- --Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU
sda4_fdc 1000 1713 99.7 15432 58.7 6202 53.0 1781 99.5 15095 47.3 126.5 4.5
In short, Raid-1 performance drops to almost 1/2 of the
performance of the non raid one device.
There is definitely something weird in raid-1 code if
others are getting same results. If you are, post them !!!
-- Mika <[EMAIL PROTECTED]>
Some usefull or useless info on my system follows...:
fs1:/# raidstart --version
raidstart v0.3d compiled for md raidtools-0.90
fs1:/# cat /proc/mdstat
Personalities : [linear] [raid0] [raid1] [raid5] read_ahead 1024 sectors
md0 : active raid5 sdc1[2] sdb1[1] sda3[0] 8192896 blocks level 5, 64k chunk,
algorithm 2 [3/3] [UUU]
md1 : active raid1 sdb2[0] sdc2[1] 4835456 blocks [2/2] [UU]
fs1:/# uname -a
Linux fs1 2.2.13ac1 #1 Sat Oct 23 18:59:06 EEST 1999 i586 unknown
fs1:/# cat /proc/scsi/scsi
Attached devices:
Host: scsi0 Channel: 00 Id: 00 Lun: 00
Vendor: WDIGTL Model: WDE9180 ULTRA2 Rev: 1.30
Type: Direct-Access ANSI SCSI revision: 02
Host: scsi0 Channel: 00 Id: 01 Lun: 00
Vendor: WDIGTL Model: WDE9180 ULTRA2 Rev: 1.30
Type: Direct-Access ANSI SCSI revision: 02
Host: scsi0 Channel: 00 Id: 02 Lun: 00
Vendor: WDIGTL Model: WDE9180 ULTRA2 Rev: 1.30
Type: Direct-Access ANSI SCSI revision: 02
fs1:/# cat /proc/scsi/aic7xxx/0
Adaptec AIC7xxx driver version: 5.1.20/3.2.4
Compile Options:
TCQ Enabled By Default : Disabled
AIC7XXX_PROC_STATS : Disabled
AIC7XXX_RESET_DELAY : 5
Adapter Configuration:
SCSI Adapter: Adaptec AIC-7880 Ultra SCSI host adapter
Ultra Wide Controller
PCI MMAPed I/O Base: 0x80000000
Adapter SEEPROM Config: SEEPROM not found, using defaults.
Adaptec SCSI BIOS: Disabled
IRQ: 9
SCBs: Active 0, Max Active 3,
Allocated 15, HW 16, Page 255
Interrupts: 893038
BIOS Control Word: 0x0000
Adapter Control Word: 0x0000
Extended Translation: Disabled
Disconnect Enable Flags: 0xffff
Ultra Enable Flags: 0x0007
Tag Queue Enable Flags: 0x0000
Ordered Queue Tag Flags: 0x0000
Default Tag Queue Depth: 8
Tagged Queue By Device array for aic7xxx host instance 0:
{255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255}
Actual queue depth per device for aic7xxx host instance 0:
{1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}
Statistics:
(scsi0:0:0:0)
Device using Wide/Sync transfers at 40.0 MByte/sec, offset 8
Transinfo settings: current(12/8/1/0), goal(12/8/1/0), user(12/15/1/0)
Total transfers 360722 (159990 reads and 200732 writes)
(scsi0:0:1:0)
Device using Wide/Sync transfers at 40.0 MByte/sec, offset 8
Transinfo settings: current(12/8/1/0), goal(12/8/1/0), user(12/15/1/0)
Total transfers 266561 (78931 reads and 187630 writes)
(scsi0:0:2:0)
Device using Wide/Sync transfers at 40.0 MByte/sec, offset 8
Transinfo settings: current(12/8/1/0), goal(12/8/1/0), user(12/15/1/0)
Total transfers 265654 (155306 reads and 110348 writes)
