Dedicated datastore for swap on NetApp...

[richard6121]

Can Windows pagefiles and VMware swap files share the same datastore? They both seem to have similar constraints regarding snapshots and replication; any reason why they can't live together?

[AWo]

The guest swapfiles reside on a guest volume, a virtual disk. That resides on a VMFS datastore. The VMware guest swap file resides only on a VMFS datastore.

If you put the guest swapfile on a RDM device, it can reside on the same LUN as the VMFS datastore, but only one guest can use that RDM device.

BTW, why do you want to keep them together? If the VMware guest swap file is used you have a problem: to less memory in the ESX host. Even if you can overcommit memory, why would you want to do that? That is a performynce killer.

AWo

VCP 3 & 4

Author @ vmwire.net

\[:o]===\[o:]

=Would you like to have this posting as a ringtone on your cell phone?=

=Send "Posting" to 911 for only $999999,99!=

[RParker]

Even if you can overcommit memory, why would you want to do that? That is a performynce killer.

AM I missing something? the pagefile livesIN the VM, it's not outside. The .vswp file is the VM swap, but ONLY comes into play if you don't size the guest OS properly..

[RParker]

Can Windows pagefiles and VMware swap files share the same datastore?

The SAN is high performance disk. Assuming you give it enough spindles, it should not be a problem. But what pagefiles are you talking about? If this is a VM datastore it should only be VM's.

[AWo]

Am I (me, AWo) missing something?

AFAIK the VMware guest swap file is used when the host lacks enough memory and VMware needs to swap out. I was referring to this as a performance killer.

From the manual regarding the SWAP file which is used when memory overcomittment has taken place and all other mechanisms (like balloning) doesn't help anymore:

Using Swap Files

You can specify the location of your swap file, reserve swap space when memory is overcommitted, and delete a swap file.

ESX/ESXi hosts use swapping to forcibly reclaim memory from a virtual machine when the vmmemctl driver is not available or is not responsive.

• It was never installed.

• It is explicitly disabled.

• It is not running (for example, while the guest operating system is booting).

• It is temporarily unable to reclaim memory quickly enough to satisfy current system demands.

• It is functioning properly, but maximum balloon size is reached.

Standard demand-paging techniques swap pages back in when the virtual machine needs them.

Note

For optimum performance, ESX/ESXi hosts use the ballooning approach (implemented by the vmmemctl driver) whenever possible. Swapping is a reliable mechanism of last resort that a host uses only when necessary to reclaim memory.

By default, the swap file is created in the same location as the virtual machine's configuration file.

So it is created on the VMFS datastore, not inside the guest.

Am I wrong?

AWo

VCP 3 & 4

Author @ vmwire.net

\[:o]===\[o:]

=Would you like to have this posting as a ringtone on your cell phone?=

=Send "Posting" to 911 for only $999999,99!=

[RParker]

So it is created on the VMFS datastore, not inside the guest.

No, you are not wrong. I am trying to figure out why there would be pagefiles (as in Windows pagefile) on the same volume / datastore as VM Swap files.

the Swap Files for VM live outside, I got that part. I am trying to figure out where the pagefile comes in, because they live in the guest, which is part of the VM.

And yes it is a performance killer, but I don't get the pagefile scenario, where did they come from?

[richard6121]

Gents,

This is a pure NFS environment. No vmfs involved.

The NetApp best practices doc shows vm's in one datastore, dedicated vmdk's for the windows pagefiles on a second datastore, and host swap on a third. Why not combine the 2nd and 3rd?

RM

[mike_laspina]

Hi Richard,

The simple answer is yes you can place the vmdk

there and it is advisable to separate noise based IO from Netapp

snapshot enabled NFS shares. However it may not be the best protocol for

that application since VMware NFS writes are all O_Sync based and this

will tax the storage servers cache. You can not control VMware NFS write

and thus the "It may not be the right protocol" comment. iSCSI or FC

would be more suitable since it can cache the entry in a more granular

way allowing the application to control if a write sync is required where in this case a page file would not be a sync based call.

Even

with that said you still need to address the noise issue so your snaps

will not hold lots of useless temp/swap file locked in snap

allocations.

Have a look at this entry

it refers to the same issues but on a ZFS system.

Regards,

Mike

vExpert 2009

[RParker]

dedicated vmdk's for the windows pagefiles on a second datastore, and host swap on a third. Why not combine the 2nd and 3rd?

AH, Got it now. It's not the PAGEFILE you are swapping it's the VMDK with the pagefile on it, now I see what they mean.

Netapp has a LOT of misinformation and confusing information. We had 3 meetings with Netapp (and this was confirmed with other people on this same community forum), each time sales, tech and engineer, they ALL conflicted each other with exactly how to configure the SAN. So you have to take what Neapp says with a grain of salt.

Also I agree, you CAN combine them with no issue, but this is the first I have seen of this configuration so it's yet another way that some other tech/engineer told you to do it.

Separating them means wasting space, since dedicating them means reserve space ONLY for that purpose, and it's extra overhead to manage.

Besides we are talking major IO to separate them like that, it seems like a complete waste to me otherwise. MOST VM's don't need or can use that much IO to warrant making them separate, and Netapp advertises their storage to be high IO, so why would we need to make it separate?

BTW we use Netapp NOW so I am VERY familiar with how to setup Volume/Aggregrate/LUN/VMFS storage.. so I know the limits, and VM's don't even approach that much IO.

[AWo]

{

but I don't get the pagefile scenario, where did they come from?

Pagefile scenario? Sorry, I can't follow you now. You quoted my post and I don't understand to wich pagefile scenario you refer to.

Can you explain, please?

AWo

VCP 3 & 4

Author @ vmwire.net

\[:o]===\[o:]

=Would you like to have this posting as a ringtone on your cell phone?=

=Send "Posting" to 911 for only $999999,99!=