Vista - Performance penalty between dynamic- and fixed-size virtual hard disks?

In some article of Microsoft's, it says a fixed-sized virtual hard disk
has better performance than a dynamically-sized virtual hard disk.
Okay, so how MUCH is the performance penalty if I use a
dynamically-sized virtual hard disk? VM's already run slower than the
host OS so the user already experiences a performance impact when
running software under the VM. That's the nature of the beast. I
haven't gotten into using Microsoft's free VM server on which I run
other guests because I don't have the hardware that could handle that
setup. So I'm using the software virtualization afforded by VirtualPC
for now.

If I create a 16GB dynamically-sized virtual hard disk, how much slower
will that guest operate regarding disk activity versus the same guest
using a fixed-size virtual hard disk? A lot? Very little? Just
noticeable?

On Tue, 23 Dec 2008 18:10:11 -0600, VanguardLH <V@xxxxxx> wrote:
The only time you'll see a difference is when the dynamic disk is
writing to the hard drive and needs to expand the disk. Reads won't
make any real difference.

Personally, if you have a fast cpu and disk subsystem, you'll probably
not notice a difference. If you have multiple disks, moving the VM's
VHD off the primary OSes drive will probably yield a more noticable
differnce than running a fixed on the same disk.

--
Cheers,
Steve Jain, Virtual Machine MVP
http://vpc.essjae.com/

Steve Jain [MVP] wrote:
I have an old single-core AMD Athlon XP 3200+ processor and 2GB RAM. I
already placed the VM files onto a separate hard drive. This because I
do incremental image backups using Acronis TrueImage and don't want the
VMs sucking up media space in the backups. If I lose the VMs, I'll
recreate them. They're just baseline OS images for software testing so
I'd only have to do the OS install, OS updates, and lockdown the
baseline VM to then do single-product software testing.

While the VM files are on a different drive, they are not on a different
disk. The first hard disk has 2 partitions: C: and D:. Acronis TI only
backups up C:. Everything on D: is already a small individual backup of
stuff on C:, like Outlook Personal Folders Backup, or where I store
downloads, the VMs, or anything else that I don't want in by Acronis
backups. The second hard disk has just one hidden partition that is
used by Acronis to store the backups. If the first hard disk dies, I
restore from the 2nd hard disk. If the 2nd hard disk dies, I still have
the first hard disk running the OS and can go get a replacement for that
2nd hard disk. C: (on 1st hard disk) gets backed up. Nothing on D: (on
1st hard disk) gets backed up. 2nd hard disk is where the backups are
stored. About 1 to 3 months I do full backups to removable media
(DVD-R) just in case both hard disks happen to get zapped. So while the
VM files are on a different drive than the OS, they are on the same disk
as the OS. I don't have a 3rd hard disk in the case yet to move the VMs
to a physically separate disk along with the pagefile, too.

The dynamically-sized virtual hard disk would already be as large as
needed to contain the OS, updates, and any support software that I
include. So you're saying write/read access to those sectors of the
..vhd file will be just as quick as if I had used a fixed-sized virtual
hard disk. Okay so far. When I do software testing, I'm not really
concerned that the install takes 5 minutes or 8 minutes, so the writes
that expand the dynamic-sized virtual hard disk being slower is okay.
Most software testing doesn't generate much, if any, data files as I'm
testing the product, not its generated documents. So once the trial
software has been installed, the dynamic-sized virtual hard disk has
been expanded and, as you say, write/read access will be just as fast as
for a fixed-size virtual hard disk because the sectors for the .vhd file
have already been allocated during the install. There might be some
file updates performed by the software during testing but, again, once
those are done and the dynamic-sized virtual hard disk has been enlarged
than any subsequent file accesses will be just as fast as with a fixed-
sized virtual hard disk.

So the only real impact that I see that I will encounter during software
testing is a slightly slower software install (which is inconsequential
to me) and first-time file creates or writes (which probably won't
matter much to me). So it looks like a dynamically-sized virtual hard
disk is the way that I'll go for now.

Thanks.

On Tue, 23 Dec 2008 19:56:01 -0600, VanguardLH <V@xxxxxx> wrote:
Yeah, once your disks expand out fully, there would be no difference
between performance, they'd both be "fixed" disks...with the option of
your dyn-ex disk having the ability to be compressed in the future.

--
Cheers,
Steve Jain, Virtual Machine MVP
http://vpc.essjae.com/

In message <mf63l4902usutv3jpr7t96e1oj8los0c8m@xxxxxx> "Steve Jain
[MVP]" <noreply.-@-.essjae.com> was claimed to have wrote:
Well, that isn't entirely true, odds are good that the expanded disk
will be heavily fragmented internally, so from a performance point of
view you're better off switching to a fixed disk.