I was told that it was best practices to let the host system manage the swap
file, however I have noticed that when I hit the physical memory limit I can
no longer start virtual machines. I was under the impression that server
2008 fixed the issue of hitting the hard limit on physical ram by swapping a
portion of it. Am I totally wrong here? If not, what would be the point to
let the host manage the swap if it does not swap to begin with?

-Confused....

On Wed, 14 Jan 2009 11:58:42 -0800, news.microsoft.com wrote:

> I was told that it was best practices to let the host system manage the swap
> file, however I have noticed that when I hit the physical memory limit I can
> no longer start virtual machines. I was under the impression that server
> 2008 fixed the issue of hitting the hard limit on physical ram by swapping a
> portion of it. Am I totally wrong here? If not, what would be the point to
> let the host manage the swap if it does not swap to begin with?

Using a pagefile for virtual memory management and over-committing memory
for virtual machines are two completely different things. No Microsoft
virtualization solution allows you to over-commit memory.

--
Paul Adare
MVP - Identity Lifecycle Manager
http://www.identit.ca

Well, coming from a VM world, I had the ability to allocate directly to
phsyical RAM or the option to allow some of the RAM for the VM Machines to
leak into the pagefile. Did I loose this functionality in Hyper-V?


"Paul Adare" <pkadare@xxxxxx> wrote in message
news:1krf86lq1sg27$.5fz9worg9cyb.dlg@xxxxxx

> On Wed, 14 Jan 2009 11:58:42 -0800, news.microsoft.com wrote:
>

>> I was told that it was best practices to let the host system manage the
>> swap
>> file, however I have noticed that when I hit the physical memory limit I
>> can
>> no longer start virtual machines. I was under the impression that server
>> 2008 fixed the issue of hitting the hard limit on physical ram by
>> swapping a
>> portion of it. Am I totally wrong here? If not, what would be the point
>> to
>> let the host manage the swap if it does not swap to begin with?

>
> Using a pagefile for virtual memory management and over-committing memory
> for virtual machines are two completely different things. No Microsoft
> virtualization solution allows you to over-commit memory.
>
> --
> Paul Adare
> MVP - Identity Lifecycle Manager
> http://www.identit.ca

On Wed, 14 Jan 2009 14:31:55 -0800, news.microsoft.com wrote:

> Well, coming from a VM world, I had the ability to allocate directly to
> phsyical RAM or the option to allow some of the RAM for the VM Machines to
> leak into the pagefile. Did I loose this functionality in Hyper-V?

The feature in VM is over-committing memory and yes, Hyper-V does not allow
that.

--
Paul Adare
MVP - Identity Lifecycle Manager
http://www.identit.ca

I somewhat disagree, because sometimes you can have a spike of memory usuage
and if you decide that due to cost swap is the best method for distributing
this spike I would not call that over commiting, just a backup plan that is
cheaper/more realistic than physical RAM. So what I gather from all of this
is that even though microsoft recomends that you do not swap in the guest
Machines, unless you have the luxury of having enough physical RAM to cover
spikes, you are better off placing a swap inside the VM's.


"Paul Adare" <pkadare@xxxxxx> wrote in message
news:1fkhrylk7of87$.ja8j22675krl.dlg@xxxxxx

> On Wed, 14 Jan 2009 14:31:55 -0800, news.microsoft.com wrote:
>

>> Well, coming from a VM world, I had the ability to allocate directly to
>> phsyical RAM or the option to allow some of the RAM for the VM Machines
>> to
>> leak into the pagefile. Did I loose this functionality in Hyper-V?

>
> The feature in VM is over-committing memory and yes, Hyper-V does not
> allow
> that.
>
> --
> Paul Adare
> MVP - Identity Lifecycle Manager
> http://www.identit.ca

>I would not call that over commiting,

That's what it's called though, it's just a term meaning you are
allocating more memory than you have for a VM and Hyper-V doesn't do
that.

>So what I gather from all of this
>is that even though microsoft recomends that you do not swap in the guest
>Machines, unless you have the luxury of having enough physical RAM to cover
>spikes, you are better off placing a swap inside the VM's.

Correct. Otherwise it's a very tight balancing act between working
set and RAM available, most of the time not worth it to do, but if
performance is the most desired option, not swapping in the VM would
be the best.

--
Bob Comer




On Thu, 15 Jan 2009 08:25:31 -0800, "news.microsoft.com"
<None@xxxxxx> wrote:

>I somewhat disagree, because sometimes you can have a spike of memory usuage
>and if you decide that due to cost swap is the best method for distributing
>this spike I would not call that over commiting, just a backup plan that is
>cheaper/more realistic than physical RAM. So what I gather from all of this
>is that even though microsoft recomends that you do not swap in the guest
>Machines, unless you have the luxury of having enough physical RAM to cover
>spikes, you are better off placing a swap inside the VM's.
>
>
>"Paul Adare" <pkadare@xxxxxx> wrote in message
>news:1fkhrylk7of87$.ja8j22675krl.dlg@xxxxxx

>> On Wed, 14 Jan 2009 14:31:55 -0800, news.microsoft.com wrote:
>>

>>> Well, coming from a VM world, I had the ability to allocate directly to
>>> phsyical RAM or the option to allow some of the RAM for the VM Machines
>>> to
>>> leak into the pagefile. Did I loose this functionality in Hyper-V?

>>
>> The feature in VM is over-committing memory and yes, Hyper-V does not
>> allow
>> that.
>>
>> --
>> Paul Adare
>> MVP - Identity Lifecycle Manager
>> http://www.identit.ca

>

Yup...Thanks for all the help!
"Robert Comer" <bobcomer-removeme-@xxxxxx> wrote in message
newstpum41a91pmje3kcdkbnf4ur82p4iakc5@xxxxxx

> >I would not call that over commiting,

>
> That's what it's called though, it's just a term meaning you are
> allocating more memory than you have for a VM and Hyper-V doesn't do
> that.
>

>>So what I gather from all of this
>>is that even though microsoft recomends that you do not swap in the guest
>>Machines, unless you have the luxury of having enough physical RAM to
>>cover
>>spikes, you are better off placing a swap inside the VM's.

>
> Correct. Otherwise it's a very tight balancing act between working
> set and RAM available, most of the time not worth it to do, but if
> performance is the most desired option, not swapping in the VM would
> be the best.
>
> --
> Bob Comer
>
>
>
>
> On Thu, 15 Jan 2009 08:25:31 -0800, "news.microsoft.com"
> <None@xxxxxx> wrote:
>

>>I somewhat disagree, because sometimes you can have a spike of memory
>>usuage
>>and if you decide that due to cost swap is the best method for
>>distributing
>>this spike I would not call that over commiting, just a backup plan that
>>is
>>cheaper/more realistic than physical RAM. So what I gather from all of
>>this
>>is that even though microsoft recomends that you do not swap in the guest
>>Machines, unless you have the luxury of having enough physical RAM to
>>cover
>>spikes, you are better off placing a swap inside the VM's.
>>
>>
>>"Paul Adare" <pkadare@xxxxxx> wrote in message
>>news:1fkhrylk7of87$.ja8j22675krl.dlg@xxxxxx

>>> On Wed, 14 Jan 2009 14:31:55 -0800, news.microsoft.com wrote:
>>>
>>>> Well, coming from a VM world, I had the ability to allocate directly to
>>>> phsyical RAM or the option to allow some of the RAM for the VM Machines
>>>> to
>>>> leak into the pagefile. Did I loose this functionality in Hyper-V?
>>>
>>> The feature in VM is over-committing memory and yes, Hyper-V does not
>>> allow
>>> that.
>>>
>>> --
>>> Paul Adare
>>> MVP - Identity Lifecycle Manager
>>> http://www.identit.ca

>>

You're welcome.

--
Bob Comer


On Thu, 15 Jan 2009 12:12:49 -0800, "news.microsoft.com"
<None@xxxxxx> wrote:

>Yup...Thanks for all the help!
>"Robert Comer" <bobcomer-removeme-@xxxxxx> wrote in message
>newstpum41a91pmje3kcdkbnf4ur82p4iakc5@xxxxxx

>> >I would not call that over commiting,

>>
>> That's what it's called though, it's just a term meaning you are
>> allocating more memory than you have for a VM and Hyper-V doesn't do
>> that.
>>

>>>So what I gather from all of this
>>>is that even though microsoft recomends that you do not swap in the guest
>>>Machines, unless you have the luxury of having enough physical RAM to
>>>cover
>>>spikes, you are better off placing a swap inside the VM's.

>>
>> Correct. Otherwise it's a very tight balancing act between working
>> set and RAM available, most of the time not worth it to do, but if
>> performance is the most desired option, not swapping in the VM would
>> be the best.
>>
>> --
>> Bob Comer
>>
>>
>>
>>
>> On Thu, 15 Jan 2009 08:25:31 -0800, "news.microsoft.com"
>> <None@xxxxxx> wrote:
>>

>>>I somewhat disagree, because sometimes you can have a spike of memory
>>>usuage
>>>and if you decide that due to cost swap is the best method for
>>>distributing
>>>this spike I would not call that over commiting, just a backup plan that
>>>is
>>>cheaper/more realistic than physical RAM. So what I gather from all of
>>>this
>>>is that even though microsoft recomends that you do not swap in the guest
>>>Machines, unless you have the luxury of having enough physical RAM to
>>>cover
>>>spikes, you are better off placing a swap inside the VM's.
>>>
>>>
>>>"Paul Adare" <pkadare@xxxxxx> wrote in message
>>>news:1fkhrylk7of87$.ja8j22675krl.dlg@xxxxxx
>>>> On Wed, 14 Jan 2009 14:31:55 -0800, news.microsoft.com wrote:
>>>>
>>>>> Well, coming from a VM world, I had the ability to allocate directly to
>>>>> phsyical RAM or the option to allow some of the RAM for the VM Machines
>>>>> to
>>>>> leak into the pagefile. Did I loose this functionality in Hyper-V?
>>>>
>>>> The feature in VM is over-committing memory and yes, Hyper-V does not
>>>> allow
>>>> that.
>>>>
>>>> --
>>>> Paul Adare
>>>> MVP - Identity Lifecycle Manager
>>>> http://www.identit.ca
>>>

>