IVM Storage and the Use of PVLinks

[Michael Patrimonio]

Greetings,

The documentation for Integrity Virtual Machine seems to contradict itself when it discusses the use of PVLinks. While it does explicitly state that PVLinks cannot be used on a VM guest for an allocated storage resource, in the description of "Attached I/O" it states that the guest VM is in complete control of the allocated LUN, which requires the use of a volume manager and, hopefully, PVLinks.

My questions are;

Is this true, can PVLinks be used with Attached I/O storage?

Are the virtual HBAs (emulated from the physical HBAs) considered storage entities, preventing the use of PVLinks on the VM guest?

If so, does my only option to use PVLinks become allocating and configuring the storage on the VM host (with LVM and PVLinks) and allocating the Logical Volumes to the VM guest?

If so, what kind of performance hit can be expected? Even a "ballpark" figure would be helpful. Are we talking about an order of magnitude?

Any information would be very much appreciated.

Thanks & Regards,
Michael
=====

[Torsten. · Accepted solution]

Hi,

if you read the docs a second time this probably becomes more clear.

You can use either a plain disk, a LVOL or a file as virtual storage.

In case of a disk device you need a multipath software for multipathing (or HP-UX 11.31 with Integrity VM once it is available).

So your option is either a LVOL or a file as a virtual disk.

In this case the *Host* system is able to use PVLINKS - the guest will use the device without knowing the alternate pathes.

Example:

If you decide to use a LVOL as virtual disk - just create a VG (with PVLINKS) with a LVOL on the *HOST* and add this lvol to the guests configuration.

If a path fails, the HOST will use the alternate path - the guest will continue to work without any notice of a failed path.

(same for a file as virtual disk)

[Michael Patrimonio]

Thank you, this is what I thought, I was just hoping to be "wrong." :-)