Comments on my idea for a Cache Drive


Recommended Posts

Ok, so in the process of planning/building my UnRAID server.

 

I was able to pick up some WD Raptors fairly cheap and thought they'd be great as my cache disk.

 

So, i have a LSI 1068E SAS/SATA card which does RAID.

 

I have 3 36GB Raptors, if i stripe them i'll get a touch over 100GB for my cache disk.

 

I understand the risks of using striping (ie, a drive dies and everything goes) and am happy with it.

 

Whats everyone think???

Link to comment

I wouldn't be comfortable with that setup.  I've had two Raptors (a 74GB and a 150GB) fail on me over the past few years.  Striping three of them and using that array for such a critical part of your system seems a bit of a risk.  Also, the heat generated by and the power consumption of three 10,000 RPM disks should be considered.

 

If it's speed your interested in, why not consider something like: http://www.microcenter.com/single_product_results.phtml?product_id=0345740

 

I realize this path would give you a smaller cache and might cost a bit more but you'd probably end up with a more dependable tower.

Link to comment

I wouldn't be comfortable with that setup.  I've had two Raptors (a 74GB and a 150GB) fail on me over the past few years.  Striping three of them and using that array for such a critical part of your system seems a bit of a risk.  Also, the heat generated by and the power consumption of three 10,000 RPM disks should be considered.

 

If it's speed your interested in, why not consider something like: http://www.microcenter.com/single_product_results.phtml?product_id=0345740

 

I realize this path would give you a smaller cache and might cost a bit more but you'd probably end up with a more dependable tower.

 

Yeah, i had thought of that...and i might go that way (the ssd) eventually. For the moment, i think i've used up my current budget for my server...i just spent over half my birthday money on a new board and RAM.

 

I have a TT Armor case and im putting the 3 raptors in the cage near the PSU, so its away from the other HDDs

 

If you mean by picking them up fairly cheaply, you got them used I would not suggest it. Unless your just doing it for the pure hobbyists sense of kicks and giggles. 

 

yea, i got them second hand...

Link to comment

I would certainly run each of those Raptors separately through 3 cycles of preclear before attempting to use them in a raid-0 mix.

 

For what it's worth, my 10Krpm 74GB Raptors individually were on par or slightly slower in write speeds than my old 7200rpm 750GB WD Black drive. If those drives are the original Raptor models, individually they will be even slower than the 74GB Raptors and so I wouldn't be surprised if even in raid-0 they end up only on-par or barely 20% faster than current-gen 7200rpm drives.

 

Here's what I experienced using an older drive write speed test. The 74GB Raptors maxed out at 78.8 MB/s at the fastest part of the disk.

WD 74GB SATA 10K RPM WD740GD-00FLA1
     sdb1 Primary Linux ReiserFS      35434.51
     sdb2 Primary Linux ext3          30721.43
     sdb3 Primary Linux swap           8192.38

/dev/sdb1# ~/diskspeed
4096000000 bytes (4.1 GB) copied, 51.9476 s, 78.8 MB/s
4096000000 bytes (4.1 GB) copied, 52.1988 s, 78.5 MB/s

/dev/sdb2# ~/diskspeed
4096000000 bytes (4.1 GB) copied, 54.9703 s, 74.5 MB/s
4096000000 bytes (4.1 GB) copied, 55.1367 s, 74.3 MB/s

 

Here's how a Seagate 2TB 5900RPM faired, notice how it performed better than the Raptor even midway through the drive, and it wasn't until the near end of the drive did the speeds drop.

Seagate 2TB SATA 5900 RPM ST32000542AS
    sdb1  Primary Linux ReiserFS        980741.27
    sdb2  Primary Linux ReiserFS        980741.27
    sdb3  Primary Linux ext3             30721.43
    sdb4  Primary Linux swap              8192.38

/mnt/tmp/sdb1# ~/diskspeed
4096000000 bytes (4.1 GB) copied, 40.8918 s, 100 MB/s
4096000000 bytes (4.1 GB) copied, 42.6702 s, 96.0 MB/s
4096000000 bytes (4.1 GB) copied, 34.6182 s, 118 MB/s

/mnt/tmp/sdb2# ~/diskspeed
4096000000 bytes (4.1 GB) copied, 49.2317 s, 83.2 MB/s
4096000000 bytes (4.1 GB) copied, 47.7500 s, 85.8 MB/s

/mnt/tmp/sdb3# ~/diskspeed
4096000000 bytes (4.1 GB) copied, 73.6467 s, 55.6 MB/s
4096000000 bytes (4.1 GB) copied, 72.5483 s, 56.5 MB/s

Link to comment

Preclear_disk.sh burns-in or exercises the disk by performing verified read and write cycles. It is extensive enough to help identify questionable drives or drives with questionable media before you put your valuable data on them only to have them die taking your data with them. Since each drive is low storage, I'd run them through 3 cycles each.

 

That diskspeed is a shell script contributed by the users on this forum. Here is the most recent version of it (Message). I now run mine with the later 3 lines uncommented so it performs the sync and reads back file so I can get an idea on read performance too.

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.