Hello Dion, and other listfolk,
Regarding your email to Adam (I didn''t want to thread-jack for the
purpose of my question):
>mount another LV somewhere on dom0 (e.g. /var/lib/xen/images/windows)
>Use "qemu-img convert your.vmdk -O raw
>/var/lib/xen/images/windows/disk0.raw" to create the file container
>I see significant disk I/O performance improvement on all my Windows
>DomU''s when running them from file containers as compared to
running
>them directly from a block device.
Is this truly the best formula you''ve determined for getting the best
performance from a Windows DomU? I know that to ask the question is a
bit like beating a dead horse, but in my time on reading the list,
browsing the website, and reading the wiki, I''ve never come up with a
clear answer, and none of my own testing has been thorough enough to
answer this myself. Further, if you know, does that recommendation
change if one isn''t using GPLPV?
Basically, should my Windows DomU storage be: block device (DAS/SAN) >
LVM volume > raw disk file > DomU? Is there a different setup for
optimal performance without GPLPV, or am I missing a step such as
mounting the image to a loopback device and then mapping the raw
loopback dev to my DomU?
I recognize that it''s possible that there is no clear answer due to
various factors, such as Xen version, Dom0 OS, kernel, and so on, and
perhaps that''s why I don''t have a good answer in my head
either. I''m
also curious if sacrificing flexibility (by omitting LVM, for example)
allows for better performance to the Windows DomU.... So many
questions :P
Thanks,
Andrew Bobulsky