›

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Sign In with Google

Sign In Register
Our Sites

covecube.com 
community.covecube.com 
blog.covecube.com 
wiki.covecube.com 
bitflock.com 
stablebit.com 

Categories

Poll

No poll attached to this discussion.

In order to better support our growing community we've set up a new more powerful forum.


The new forum is at: http://community.covecube.com


The new forum is running IP.Board and will be our primary forum from now on.


This forum is being retired, but will remain online indefinitely in order to preserve its contents. This forum is now read only.


Thank you,


Read/Write performance

eagleknighteagleknight
edited September 2011 in DrivePool
I know M4 is going to concentrate on the performance end of DrivePool. I was just curious what speeds other people are currently getting when copying files. I only am getting around 7-11MB/s, which seems very slow when your used to gigabit.
Share on Twitter

Comments

  • SilverRubiconSilverRubicon
    Member
    Copying a 900mb file to a non duplicated share I average 60-70mb/s.  To a duplicated share I average 10-20mb/s.  It is much slower, but I don't write to duplicated shares all that often.  Would be nice to only see a 50% decrease in performance.
    Share on Twitter
  • eagleknighteagleknight
    Member
    I didn't think of that between the duplicated and non-duplicated folder before you said it, but yet on non duplicated ones I see around 70 as well.
    Share on Twitter
  • AlexAlex
    edited September 2011 Covecube
    I'm testing an internal build of M4 with optimized read I/O. I'm seeing up to 90MB / s read time, but your milage may vary.

    M4 uses direct driver to driver I/O in the kernel, it also takes advantage of the NT cache manager (called Fast I/O in the kernel).

    In addition, the current build of M4 has an experimental striping feature. Basically, duplicated folders are read across 2 disks at the same time combining the read speeds of both disks.

    But things still depend on other factors like what else the server is doing at the time, bus speed to the disks (USB 2.0 is slow), the speed of the disks themselves and the speed of the CPU. So as I said, your milage may vary.

    But I think we can say goodbye to 10 MB/s in M4.

    With time I'll put up more concrete numbers.
    Share on Twitter
  • SilverRubiconSilverRubicon
    Member
    All of that sounds fantastic but also dangerous (or scary, select the adjective of your choice).  Will we need to be extra vigilant with M4 as far as data consistency goes?
    Share on Twitter
  • eagleknighteagleknight
    edited September 2011 Member
    If you are referring to the read across both disks Alex mention. That wouldn't be dangerous. The data is already on both disks. It would be reading from both disks to increase the throughput.
    Share on Twitter
  • AlexAlex
    Covecube
    @SilverRubicon: I don't think it will be any worse than M3 in terms of stability. I'm actually hopeful that it will be more reliable than what it's doing now.
    Share on Twitter
  • peste19peste19
    Member
    cant wait for m4 to come out to get the extra performance
    Share on Twitter
  • JohnBrahamJohnBraham
    Member
    That's great news about the performance, I've stopped using M3 for now as the performance was too bad, such as browsing directories with many high resolution images (something the wife does daily), but I thought it was just me, I look forward to M4, I'll re-install drivepool when that appears.
    Share on Twitter
Sign In or Register to comment.