PE 1800 PERC4/SC write performance

Irwan Hadi ihblist at gmail.com
Sat Apr 29 02:16:12 CDT 2006


This box runs Redhat ES 4 Update 2 (haven't updated it to update 3
yet), kernel is stock from ES 4 including megaraid driver.
CPU is single P IV Xeon 2.8 Ghz 1 MB cache, with 1 GB memory and DRAC card.

PERC firmware currently is 521S (I don't see any reason to upgrade it
to 521X, since the server doesn't serve any RAID 5). The RAID volume
setting is also default, 64 K stripe size, adaptive read ahead, write
back.

The hard drive is Maxtor U320. Though I doubt about this, but I got a
feeling that PERC controller seems to run better with Maxtor hard
drives compared to Seagate. Problem it is, when you order a server
from Dell, or getting a replacement hard drive, Dell may send you
either Maxtor, Seagate, or Hitachi.

What is the brand of hard drive you used in your RAID volume?

Anyway, I just re-run the iozone test again as attached. You may want
to download iozone, and run it with the same parameter I used below to
get an apple to apple comparison.
Please note that the test I did with iozone is sequential read/write.
I did not do random read/write which should be more interesting. I'll
try to run the random test tomorrow if I have time.

# cat /etc/redhat-release
Red Hat Enterprise Linux ES release 4 (Nahant Update 2)
#

# uname -r
2.6.9-22.0.2.ELsmp
#


# dmesg | grep megaraid
megaraid cmm: 2.20.2.6 (Release Date: Mon Mar 7 00:01:03 EST 2005)
megaraid: 2.20.4.6 (Release Date: Mon Mar 07 12:27:22 EST 2005)
megaraid: probe new device 0x1028:0x0013:0x1028:0x016c: bus 2:slot 14:func 0
megaraid: fw version:[521S] bios version:[H430]
#

        Iozone: Performance Test of File I/O
                Version $Revision: 3.263 $
                Compiled for 32 bit mode.
                Build: linux

        Contributors:William Norcott, Don Capps, Isom Crawford, Kirby Collins
                     Al Slater, Scott Rhine, Mike Wisner, Ken Goss
                     Steve Landherr, Brad Smith, Mark Kelly, Dr. Alain CYR,
                     Randy Dunlap, Mark Montague, Dan Million,
                     Jean-Marc Zucconi, Jeff Blomberg,
                     Erik Habbinga, Kris Strecker, Walter Wong.

        Run began: Fri Apr 28 23:17:25 2006

        Using minimum file size of 524288 kilobytes.
        Using maximum file size of 2097152 kilobytes.
        Auto Mode
        Cross over of record size disabled.
        Command line used: ./iozone -n 512m -g 2g -f test.out -az -i0 -i1
        Output is in Kbytes/sec
        Time Resolution = 0.000001 seconds.
        Processor cache size set to 1024 Kbytes.
        Processor cache line size set to 32 bytes.
        File stride size set to 17 * record size.
                                                            random 
random    bkwd  record  stride
              KB  reclen   write rewrite    read    reread    read  
write    read rewrite    read   fwrite frewrite   fread  freread
          524288       4  116317  113808  1772703  1840292
          524288       8  114854  120219  2065068  2225981
          524288      16  111525  130206  2559287  2607152
          524288      32  112515  121775  2783952  2811194
          524288      64  103835  119138  2898989  2952238
          524288     128  100199  130411  2774317  2989883
          524288     256  112328  113535  2670223  2706604
          524288     512  110654  118731  1455760  1456634
          524288    1024  105159  117243   973331   973031
          524288    2048  113879  112757   965580   968117
          524288    4096  100692  117923   958543   967047
          524288    8192  111236  116958   961931   968190
          524288   16384   99218  117678   972988   971187
         1048576       4   43677   61637    37791    37952
         1048576       8   58977   42134    40214    38473
         1048576      16   57517   58327    36692    39319
         1048576      32   56170   61008    40561    40168
         1048576      64   56747   47200    39596    38872
         1048576     128   56324   60849    39935    41514
         1048576     256   56840   41751    38848    39618
         1048576     512   50907   60573    40856    40253
         1048576    1024   53523   62029    40918    40133
         1048576    2048   55888   44230    39997    40382
         1048576    4096   56409   59793    41905    40554
         1048576    8192   55740   59926    39900    42858
         1048576   16384   57921   64732    40517    40578
         2097152       4   46289   41911    37264    37428
         2097152       8   41758   47316    37360    37212
         2097152      16   41530   43384    37267    37116
         2097152      32   41321   47728    37173    36874
         2097152      64   42442   45038    37268    37237
         2097152     128   41348   45354    37016    37296
         2097152     256   42604   43768    37317    37273
         2097152     512   44324   42147    37650    37429
         2097152    1024   47784   48378    38328    38085
         2097152    2048   44359   47817    38368    37935
         2097152    4096   41916   47028    37932    38055
         2097152    8192   42358   47418    37626    38000
         2097152   16384   47218   48987    38311    37649

iozone test complete.




On 4/28/06, Matthew DeJong <mdejong at dejonginc.com> wrote:
> Irwan, can you give more details about your configuration on the PE 1850
> with Perc 4e/Si? I am interested in anything from the typical
> installation of RHEL 4 that you turned off purposefully did not include.
> Which drivers you are using. Also, Bios settings for the Perc controller
> and if you made any particular tweaks. I have been reading a lot about
> low/poor write performance on the Perc 4 series cards on RHEL 3 & 4, and
> yours is the first setup that I have come across on the list that shows
> a more acceptable RW speed.
>
> I have 3 PE 2850's with PERC 4e/DI RAID 1 with 2 73 GB 10k rpm drives
> and dual 3.0 GHz Zeon processor (32bit).  Probably not pertinent, but I
> also have Adaptec 39160 cards in each.  I am running RHEL 4 U 2 and my
> write performance needs to be improved.  Here is an example Bonnie
> output which tends to be very similar for 100MB, 500MB, 1GB:
>
>               -------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
>           200 39086 97.4 218117 75.5 541264 94.3 46675 98.8 1883426 97.5
> 85235.1 161.9
>
> This performance is happening whether or not the boxes are under load.
> I am new to linux so hopefully I can provide similar information to what
> I have been seeing.  I have been playing with the installation of Dell
> megaraid drivers and testing difference Bios versions as well, so here
> some information that might help.
>
> Kernel: 2.6.9-22.ELsmp
>
> Box 1 & 2
> megaraid cmm: 2.20.2.5 (Release Date: Fri Jan 21 00:01:03 EST 2005)
> megaraid: 2.20.4.4 (Release Date: Thu Jan 27 00:01:03 EST 2005)
> megaraid: fw version:[521S] bios version:[H430]
> megaraid: probe new device 0x1028:0x0013:0x1028:0x016d: bus 2:slot
> 14:func 0
>
> Box 3
> megaraid cmm: 2.20.2.5 (Release Date: Fri Jan 21 00:01:03 EST 2005)
> megaraid cmm: 2.20.2.6 (Release Date: Mon Mar 7 00:01:03 EST 2005)
> megaraid: 2.20.4.4 (Release Date: Thu Jan 27 00:01:03 EST 2005)
> megaraid: 2.20.4.6 (Release Date: Mon Mar 07 12:27:22 EST 2005)
> megaraid: fw version:[521X] bios version:[H430]
>
> Thanks in advance for your help.
>
> Matt DeJong
>
> _______________________________________________
> Linux-PowerEdge mailing list
> Linux-PowerEdge at dell.com
> http://lists.us.dell.com/mailman/listinfo/linux-poweredge
> Please read the FAQ at http://lists.us.dell.com/faq
>



More information about the Linux-PowerEdge mailing list