On Thu, 15 Mar 2012, Simone Ricci wrote:
On Thu, Mar 15, 2012 at 3:21 PM, Bob Friesenhahn
<[email protected]> wrote:
The reads likely indicate that 4GB of RAM is not nearly enough. If the data
can not be cached in RAM, then it needs to be re-read from disk.
Yes, that's quite clear. But why it needs to READ from the pool while
I'm only writing to the ZVOL (with dd)?
If your write requests do not perfectly overlap the underlying zfs
blocks, then zfs may need to read the existing block and modify it
with your changed data prior to writing it to a new place on disk.
If the server has enough RAM, then it is likely that the existing data
is already cached in RAM so that zfs can do the update and write
without needing to access the disk first. Zfs loves RAM.
It is useful to make sure that the zfs volume blocksize matches the
I/O size which is actually used, or is at least small enough that
there is not substantial read/write "amplification".
Bob
--
Bob Friesenhahn
[email protected], http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer, http://www.GraphicsMagick.org/
-------------------------------------------
illumos-discuss
Archives: https://www.listbox.com/member/archive/182180/=now
RSS Feed: https://www.listbox.com/member/archive/rss/182180/21175430-2e6923be
Modify Your Subscription:
https://www.listbox.com/member/?member_id=21175430&id_secret=21175430-6a77cda4
Powered by Listbox: http://www.listbox.com
