> Hello !
>
> I'm using VPC2004 to duplicate some of my client's production environments
> to test applications I'm developing. Each of these virtual machine uses a
> single dynamic VHD.
>
> From time to time, I chkdsk/defrag/compress the VHD to regain some (real)
> disk space.
>
> Lately, I noticed a sensible performance degradation while running virtual
> machines, and found the usual maintenance did not improved it
> significantly.
>
> Even after chkdsk/defragging the host disk, I felt the actual performance
> improvement was not as much as I expected.
>
> Am I correct to think the guest disk maintenance operations may have a
> reverse (or compound) effect on the host disk -- ex:
> defragging/compressing the guest disk causes a larger host .vhd file
> fragmentation ?
>
> Would using a fixed size .vhd minimize/prevent this host disk
> fragmentation ?
>
> Is there a particular order (host/guest) the maintenance should be done ?
>
> Any other thoughts about this ?
>
> Thanks..
>
>

> To my knowledge, there are three types of *fragmentation* (not sure of the
> exact word here) a dynamically resizing VHD can suffer. The first one if the
> fragmentation of the vhd file on the host, the second one is the
> fragmentation of the files of the virtualized OS in the guest and the third
> one is the physical ordering of the blocks (size of 2Meg each) in the VHD
> file itself.
>
> You should defrag/compress the VHD in the guest first and then on the host
> after but this will get you rid of only the first and second types of
> fragmentation and not of the third one. The effect of the third one is
> probably what you are seeing now. The only way of getting rid of this third
> one is probably to clone your VHD to another VHD using a disk imaging
> software like Ghost or Acronis (they are many other on the market).
>
> BTW, using an imaging disk software is probably the fatest and efficient way
> of compacting/defragging a VHD in a single step. The only problem is that
> the serial number of the drive will change.
>
> Beside using an imaging disk software, the second solution would be to use a
> fixed size VHD but like you said, you will lose some hard disk space on the
> host. This is a trading here: speed and convenience against some lost of
> space on the host.
>
> --
> Sylvain Lafontaine, ing.
> MVP - Technologies Virtual-PC
> E-mail: sylvain aei ca (fill the blanks, no spam please)
>
> "Michel S." <(email address - cut out)> wrote in message
> news:(email address - cut out)...
>> Hello !
>>
>> I'm using VPC2004 to duplicate some of my client's production environments
>> to test applications I'm developing. Each of these virtual machine uses a
>> single dynamic VHD.
>>
>> From time to time, I chkdsk/defrag/compress the VHD to regain some (real)
>> disk space.
>>
>> Lately, I noticed a sensible performance degradation while running virtual
>> machines, and found the usual maintenance did not improved it
>> significantly.
>>
>> Even after chkdsk/defragging the host disk, I felt the actual performance
>> improvement was not as much as I expected.
>>
>> Am I correct to think the guest disk maintenance operations may have a
>> reverse (or compound) effect on the host disk -- ex: defragging/compressing
>> the guest disk causes a larger host .vhd file fragmentation ?
>>
>> Would using a fixed size .vhd minimize/prevent this host disk fragmentation
>> ?
>>
>> Is there a particular order (host/guest) the maintenance should be done ?
>>
>> Any other thoughts about this ?
>>
>> Thanks..
>>
>>

> Merci Sylvain !
>
> Seems the issue is still unclear, because a little before I read your
> reply, I found an article on Michael Coates blog where he suggest to
> defragment the host first - upon many other tricks he gives to optimize a
> VPC image. (http://blog.opsan.com/archive/2005/05/10/462.aspx)
>
> On the "fixed size" vhd side, do you think it remains at the same location
> on the host disk once cretated, (like a paging file) or if it's location
> can move as it contents is changing ?
>
> MS
>
>
> Dans son message précédent, Sylvain Lafontaine a écrit :
>> To my knowledge, there are three types of *fragmentation* (not sure of
>> the exact word here) a dynamically resizing VHD can suffer. The first
>> one if the fragmentation of the vhd file on the host, the second one is
>> the fragmentation of the files of the virtualized OS in the guest and the
>> third one is the physical ordering of the blocks (size of 2Meg each) in
>> the VHD file itself.
>>
>> You should defrag/compress the VHD in the guest first and then on the
>> host after but this will get you rid of only the first and second types
>> of fragmentation and not of the third one. The effect of the third one
>> is probably what you are seeing now. The only way of getting rid of this
>> third one is probably to clone your VHD to another VHD using a disk
>> imaging software like Ghost or Acronis (they are many other on the
>> market).
>>
>> BTW, using an imaging disk software is probably the fatest and efficient
>> way of compacting/defragging a VHD in a single step. The only problem is
>> that the serial number of the drive will change.
>>
>> Beside using an imaging disk software, the second solution would be to
>> use a fixed size VHD but like you said, you will lose some hard disk
>> space on the host. This is a trading here: speed and convenience against
>> some lost of space on the host.
>>
>> --
>> Sylvain Lafontaine, ing.
>> MVP - Technologies Virtual-PC
>> E-mail: sylvain aei ca (fill the blanks, no spam please)
>>
>> "Michel S." <(email address - cut out)> wrote in message
>> news:(email address - cut out)...
>>> Hello !
>>>
>>> I'm using VPC2004 to duplicate some of my client's production
>>> environments to test applications I'm developing. Each of these virtual
>>> machine uses a single dynamic VHD.
>>>
>>> From time to time, I chkdsk/defrag/compress the VHD to regain some
>>> (real) disk space.
>>>
>>> Lately, I noticed a sensible performance degradation while running
>>> virtual machines, and found the usual maintenance did not improved it
>>> significantly.
>>>
>>> Even after chkdsk/defragging the host disk, I felt the actual
>>> performance improvement was not as much as I expected.
>>>
>>> Am I correct to think the guest disk maintenance operations may have a
>>> reverse (or compound) effect on the host disk -- ex:
>>> defragging/compressing the guest disk causes a larger host .vhd file
>>> fragmentation ?
>>>
>>> Would using a fixed size .vhd minimize/prevent this host disk
>>> fragmentation ?
>>>
>>> Is there a particular order (host/guest) the maintenance should be done
>>> ?
>>>
>>> Any other thoughts about this ?
>>>
>>> Thanks..
>>>
>>>
>
>