Hi,
I have attached some quick bonnies I made on a 2x450PIII, 510MB memory,
2x3 IBM-DMVS18V SCSI disks (U2W-LVD/SE) on the two channels of a Adaptec
AHA-394X. It is a standard 2.2.13 kernel with the raid-0.9 patches
included in debian potato (August 99). No special tuning except what is
written in the software howto. The first bonnie is non-raid to see what a
disk alone can do. The raid0 on two disks has about 42MB/s for block read.
I'd say it will be cpu-bound for 3 or 4 disks. Looking at the numbers for
the other raid-levels I thought the behaviour is just as one would expect.
I will probably repeat the tests with greater files to minimaze cache
usage.
-Peter
On 03-Feb-2000 Glenn McGrath wrote:
>
>> 2. We have been testing s/w RAID for more than 8 months in both SCSI
>> and
>> FC systems
>> and have found some issues which we intend to fix post haste. The most
>> significant
>> is performance. We have yet to get s/w RAID to perform any faster than
>> 25MB/s
>> (according to Bonnie) in any configuration or RAID level. When my test
>> server
>> comes back from the LinuxWorld show I will post the numbers I'm talking
>> about and
>> perhaps you will offer some answers to this problem.
>>
>> There are some bugs which we have fixed and we will be posting a patch
>> soon to fix
>> them. These mostly relate to hot-swapping drives and failure handling.
>
> 25MB/s ?
> Hmm, i find this very surprising, you should be able to get this
> performance with 2 ide drives in raid0 mode.
>
> A DMA/33 ide drive can get 17MB/s or so by itself (ive heard reports of
> 20MB/s).
>
> Ive personally gotten 37MB/s with 3 quantum ide drives in raid0 mode,
> each drive on its own channel.
>
> When i added a fourth drive i started getting irq timeout problems due
> to my hpt366 controller (i think), on a small test i got around 39MB/s
> but i dont really trust that result.
>
> Personally i think theres a bottleneck around 40MB/s, but i could be
> wrong.
>
> I think you should be able to get better than 25MB/s without too much
> problem.
>
> Mind you i only play with ide, i dont see a need for scsi unless you
> need more than 4 devices or so. (not that i want to start a scsi vs ide
> debate) From what ive heard from other people you have to be carefull
> not to saturate the scsi bus with more than a few scsi drives.
>
> Thanks
>
> Glenn McGrath
>
-------------------------------
Email: [EMAIL PROTECTED]
WWW: http://www.risc.uni-linz.ac.at/people/ppregler
Version 0.99b ------Sequential Output------ --Sequential Input- --Random-
-Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine MB K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP
sdb1 1024 7175 97 29991 30 12699 31 7218 94 26967 31 826.2 11
------Sequential Create------ --------Random Create--------
-Create-- --Stat--- -Delete-- -Create-- --Stat--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP
30 147 99 545 99 5832 98 150 99 687 99 589 83
###########################################################################
raid 1 (sda/sdd):
Version 0.99c ------Sequential Output------ --Sequential Input- --Random-
-Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine MB K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP
Unknown 1024 7260 99 24888 28 8783 22 7438 98 16650 22 739.2 12
------Sequential Create------ --------Random Create--------
-Create-- --Stat--- -Delete-- -Create-- --Stat--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP
30 168 99 564 99 5938 99 175 99 766 99 618 79
Unknown,1024,7260,99,24888,28,8783,22,7438,98,16650,22,739.2,12,30,168,99,564,99
,5938,99,175,99,766,99,618,79
###########################################################################
raid 0 (sda/sdd):
Version 0.99c ------Sequential Output------ --Sequential Input- --Random-
-Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine MB K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP
Unknown 1024 7404 99 54473 55 18408 46 7350 96 42609 54 831.9 11
------Sequential Create------ --------Random Create--------
-Create-- --Stat--- -Delete-- -Create-- --Stat--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP
30 170 99 566 99 6022 99 181 99 794 100 740 91
Unknown,1024,7404,99,54473,55,18408,46,7350,96,42609,54,831.9,11,30,170,99,566,9
9,6022,99,181,99,794,100,740,91
###########################################################################
raid 5 (sdb/sdc/sde/sdf)
Version 0.99c ------Sequential Output------ --Sequential Input- --Random-
-Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine MB K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP
Unknown 1024 6590 95 37995 53 16473 54 7154 97 31310 48 796.0 14
------Sequential Create------ --------Random Create--------
-Create-- --Stat--- -Delete-- -Create-- --Stat--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP
30 164 99 550 99 5910 99 183 99 830 100 786 94
Unknown,1024,6590,95,37995,53,16473,54,7154,97,31310,48,796.0,14,30,164,99,550,9
###########################################################################
raid 5 (/sd[a-f])
Version 0.99e ------Sequential Output------ --Sequential Input- --Random-
-Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine MB K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP
6xraid5 1024 6609 94 42885 74 19455 65 7106 97 38243 53 18.7 0
------Sequential Create------ --------Random Create--------
-Create-- --Stat--- -Delete-- -Create-- --Stat--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP
30 170 98 578 99 6111 99 180 99 754 99 767 95
