[zfs-discuss] ZFS on RAM optimizations questions
durval.menezes at gmail.com
Fri Oct 18 21:24:38 EDT 2013
I think you are just subjecting ZFS to inhuman and cruel treatment ;-)
Seriously, ZFS was not made for that.
Just for curiosity's sake, have you tried measuring performance with a more
"traditional" ramdisk file system like tmpfs or even zram?
On Fri, Oct 18, 2013 at 9:35 PM, <maxpil at gmail.com> wrote:
> I am running ZFS on RAM disk using brd module (/dev/ram0). The RAM disk is
> Here is my config:
> zpool create -O atime=off -O dedup=on -O sync=disabled mypool /dev/ram0
> the performance I am getting is about 186MB/s sequential write. I was
> thinking I should be getting more than this writing to RAM with ZIL
> I have also tried this config:
> zpool create -O atime=off -O dedup=on -O primarycache=none -O
> secondarycache=none -O sync=disabled mypool /dev/ram0
> This config resulted in much worse performance.
> - Should primarycache/secondary cache be disabled on RAM disk? If so, why
> did I see worse performance with them disabled
> - what other RAM disk block devices I can try in Linux? I am aware of zram
> and rapiddisk
> - what module options would be helpful to optimize for RAM disk?
> - should I limit ARC, and if so how much?
> - is there still a per VDEV IOPS limit? I only have 1 RAM device in the
> pool that is 150GB would it make sense to do a stripe across several RAM
> To unsubscribe from this group and stop receiving emails from it, send an
> email to zfs-discuss+unsubscribe at zfsonlinux.org.
To unsubscribe from this group and stop receiving emails from it, send an email to zfs-discuss+unsubscribe at zfsonlinux.org.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the zfs-discuss