[linux-lvm] about snapshot performance
mauelshagen at redhat.com
Sat Nov 19 10:10:23 UTC 2005
On Sat, Nov 19, 2005 at 05:57:04PM +0800, Lee Rocky wrote:
> I found some reference that discuss about LVM1 volume with snapshot performance
> and I follow the steps with my LVM2
> "sync ; time dd if=/dev/zero of=filename bs=1024k count=2000"
> the result is
> dd write on LV without snapshots = 150MB/s
> dd write on LV with 1 snapshots = 4.42 MB/s
> I got very slow write performace.
> but read has no problem.
> is this result work as expected?
Hrm, this should perform better.
Do you have both origin and snapshot allocated to the same PV ?
It will slow down writes, because on every change to the origin,
an exception takes place and the original content needs to be preserved
(causing additional reads and writes) before the update write is allowed
to go through.
If both go to the same spindle it'll explain what you see.
If this is true, try seperating allocation of the snapshot from
the origin (ie. add PVPath to the snapshot create command).
Of course you need at least 2 PVs in your VG to do that and enough
free space on the second to allocate the snapshot.
> KERNEL 2.6.13
> LVM2 = 2.01.15
> Library : 1.01.05
> Driver version: 4.4.0
> linux-lvm mailing list
> linux-lvm at redhat.com
> read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/
Heinz -- The LVM Guy --
*** Software bugs are stupid.
Nevertheless it needs not so stupid people to solve them ***
Heinz Mauelshagen Red Hat GmbH
Consulting Development Engineer Am Sonnenhang 11
Cluster and Storage Development 56242 Marienrachdorf
Mauelshagen at RedHat.com +49 2626 141200
More information about the linux-lvm