Take the 2-minute tour ×
Ask Different is a question and answer site for power users of Apple hardware and software. It's 100% free, no registration required.

I just bought a LaCie Rugged Safe Hard drive, and I am a bit disappointed by the sustained write speed that I measured.

Here is what I see :

  • sustained read speed : about 55MB/s
  • sustained write speed about 12MB/s if the disk is formatted with HFS/journaled
  • sustained write speed about 20MB/s if the disk is formatted with HFS/Not journaled
  • sustained write speed about 55MB/s if the disk is formatted with HFS/Not journaled, but I write to it through a Parallels Desktop virtual machine !!!

Does anyone have a clue on a possible reason why the sustained write speed would be faster from a virtual machine?

Also, does anyone what I should do to have the same write speed from Mac OS?

Technical info

  • I do not need to connect the disk as a Virtual Machine device, I need only to access it as a disk that is shared by the host OS
  • The disk is connected via a firewire 800 cable
  • It is the most recent LaCie Rugged Safe 1Go drive
  • Here is the output of a mount command
$mount
/dev/disk2s2 on /Volumes/LaCie-Rugged (hfs, local, nodev, nosuid, noowners)
share|improve this question
    
Your speeds are normal for that drive. What sort of data is being written to test the write speeds? Usually test data is compressible and cacheing or zero compression effects are why some VM report faster IO. They get rid of the data and the mac still writes is using the best speed it can. The only difference could be writing to block or raw device and skipping spotlight indexing. Have you tried asking spotlight to not index that drive by listing it in the privacy preference pane under spotlight? –  bmike Jun 15 '11 at 22:48
    
Well, I did a test by copying 3GB of compressed movies, so the data is mainly not compressible. I did not deactivate spotlight, but the disk was blank during my test, so I suspect that spotlight is out of cause. –  Pascal T. Jun 16 '11 at 22:20
    
I just did the same test with spotlight deactivated, and I still have the same speed –  Pascal T. Jun 16 '11 at 22:28
    
Bummers. mds and mdworker do process all writes so I was hoping you could eliminate them as a cause. Looks like you did and it wasn't the cause. Unless there's some weird interaction between the source disk for your 3GB files, I can't think of anything you can tune at this point. –  bmike Jun 16 '11 at 23:34
add comment

1 Answer

I would be wary of accepting any sustained write speed test results from inside a VM as fact. Most VM architectures execute clever caching and other algorithms when writing to virtual disks. Try running the "dd if=/dev/zero of=/dev/sda bs=1024k" command from inside an empty Linux VM against a large virtual disk to see exactly what I mean (speeds far higher than your physical storage).

How are you testing the firewire drive from within the VM? Is the drive being passed through to the VM as a firewire device in Windows or as some other device/drive?

More details would be very much appreciated.

share|improve this answer
    
The device is connected to the VM as a network drive, shared by the host OS (i.e I can access it as \\psf\LaCie-Rugged) ! I am copying compressed data (movies), so I think that it is safe to say that no data compression is possible during the transfer. I just did the same test again, with 3GB of movies : 52MB/s on the VM, 26MB/s on OSX ! –  Pascal T. Jun 16 '11 at 22:24
1  
@Pascal T. - It may not be doing compression, but it could be doing a fair amount of other things: deferred writes, etc... –  Fake Name Sep 13 '11 at 11:09
    
Additionally being a hybrid kernel instead of a micro kernel operation system, Windows IO is generally faster than the OS X IO. –  Max Ried Feb 14 '12 at 12:21
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.