Topic: Anyone ever tired this... 2 SATA drives and 1 PATA...

[harshl]

I am going to be standing up a new SME 7 server for use at home. I will be purchasing 2 Seagate ST3320620AS 320GB drives and 1 Seagate ST3320620A. The only difference between these drives is that one is SATA and the other is PATA.
Anyone mixed and matched SATA and PATA drives in an SME software RAID config?
I am going to test mixing SCSI and IDE in VMWare this weekend, but this is hardly an accurate test for what I will be doing.

Let me know if you have had similar experiences that you can share.

Thanks,
-Landon

[jfarschman]

Landon,

  I don't think mixing drives with differing throughput is a very good idea.  RAID is going to cause it all to work at the speed of the slowest drive.  I think you'll see noticible performance slowdowns.

  Have you considered just using the two SATAs in the mirror and installing the PATA afterward.  Then you'd have another volume attached.  Pretty nice.

[harshl]

That is not a bad idea, and I will consider it, but redundancy was as important as storage. I would actually be surprised if I saw much slow down putting an PATA dirve in the mix with SATA. SATA has a much higher theoretical bandwidth (in this case 150 MBps) however I will be limited by the speed of the drive. I may have a burst here and there get throttled by the PATA bus, however, these drives will not sustain transfer rates over 100MB anyway so again, I think it will be a non-issue. Just curious if the software raid will panic in a configuration like this.

Thanks for the reply anyway.

Anyone out there ever tried this?

Thanks,
-Landon

[paquerette]

I agree with you Harshi, throughput are not so different in this case (the 3 drives a recent) and won't be a problem here.

You can put any block device in a raid, it's advised to have almost identical drives (which is the case for you)


Also I've made an install with this configuration :

1 old PATA 120GB drive
4 new SATA 250GB drives

and by default the SME creates /dev/md1 using each 5 drives, raid 1 level, with no issue.

So go on

Paquerette

[harshl]

Thank you for the reply paquerette!

I will give it a shot.

-Landon

[Xlot]

I'm using 3x 160Gb Seagate drives.. 2 are SATA, 1 is IDE.

All works fine - they're all 8Mb cache drives of the same generation (barracuda 7200.9) so identical except for the interface.

Code: [Select]

mdadm -Q --detail /dev/md2
/dev/md2:
        Version : 00.90.01
  Creation Time : Sat Jul 22 11:57:49 2006
     Raid Level : raid5
     Array Size : 312367616 (297.90 GiB 319.86 GB)
    Device Size : 156183808 (148.95 GiB 159.93 GB)
   Raid Devices : 3
  Total Devices : 3
Preferred Minor : 2
    Persistence : Superblock is persistent

    Update Time : Tue Jul 25 18:47:51 2006
          State : clean
 Active Devices : 3
Working Devices : 3
 Failed Devices : 0
  Spare Devices : 0

         Layout : left-symmetric
     Chunk Size : 256K

    Number   Major   Minor   RaidDevice State
       0       3        2        0      active sync   /dev/hda2
       1       8        2        1      active sync   /dev/sda2
       2       8       18        2      active sync   /dev/sdb2
           UUID : f2f1fdf7:59d064b7:8117eb30:f47a9799
         Events : 0.256058


[Xlot]

Oh, and in case anyone else is interested, here's hdparm -t output for the array, and also one drive of each type.

As you can see, not a lot of difference in performance between IDE and SATA (although it is a consistent difference over multiple timings - some variation, but the SATA seems to average 1.5MB/sec higher rate).  And the software raid is decent - not stellar, but I'm in it for the R part of RAID

Code: [Select]


hdparm -t /dev/md2

/dev/md2:
 Timing buffered disk reads:  186 MB in  3.01 seconds =  61.70 MB/sec
# hdparm -t /dev/sdb2

/dev/sdb2:
 Timing buffered disk reads:  198 MB in  3.01 seconds =  65.73 MB/sec
# hdparm -t /dev/hda2

/dev/hda2:
 Timing buffered disk reads:  194 MB in  3.03 seconds =  64.06 MB/sec



[harshl]

This is great information, thank you!