Use case for NPIV?

Essendon

Been reading up on NPIV and I dont see the point of having multiple WWPN's in your VM's. You still need to add the RDM disk to the VM's, so why go through the extra work of configuring NPIV (zoning and generating the WWPN's) ? What benefit does it provide?

Someone please shed some light on this?

dave330i

Scott Lowe has a nice blog about it: Understanding NPIV and NPV - blog.scottlowe.org - The weblog of an IT pro specializing in virtualization, storage, and servers

Use case is very limited.

Essendon

Yeah that was one of the first blogs I read for NPIV. Looks like it's a fledgling technology and perhaps VMware can give us something more useful to do with it down the track. Like someone was saying over on Cormac Hogan's blog that a virtual FC HBA card would be nice.

jmritenour

What makes it useful is the fact that you can put a VM in it's own independent zone on your fabric. That way, you can present storage directly to a VM, and not the hypervisor itself, so only that VM can touch it. Granted, it's pretty hard to end up with a scenario in which you have a RDM in a trespass situation, but zoning directly to the VM would mitigate that almost entirely.

Really, the only time I found it useful was with clustering within the guest OS.

Essendon

Thanks for the post Jas, I understand that you can put the VM in its own zone and present storage directly. But an RDM is the same thing too, you give the VM exclusive access to the LUN and your bypassing the hypervisor. I guess I'm after performance differences between the two methods and if one is better than the other in certain situations.

blargoe

I really don't see much of a point in using it in its current form, but maybe they are going to build on the feature in a later release. Maybe if you were using it to do more granular, per VM tracking of metrics on your storage controller or San switch, but even then, if you have a decent vCenter integrated monitoring package, you are going to see IO issues and find the correlation without going through the complexity of setting up NPIV for the VMs.

jmritenour

Essendon wrote: »

Thanks for the post Jas, I understand that you can put the VM in its own zone and present storage directly. But an RDM is the same thing too, you give the VM exclusive access to the LUN and your bypassing the hypervisor. I guess I'm after performance differences between the two methods and if one is better than the other in certain situations.

Exactly, there is no performance difference or any other benefit with NPIV vs a plain old RDM - it's almost entirely for zoning purposes and visibility at the storage level as to what is actually using space.

QHalo

nm, still sleepy

down77

Take a look at the follow blog post. While you may not see much of a performance differential using NPV/NPIV as opposed to RDM, there are a number of important business and compliance considerations that this can address.

N_Port Virtualization with vSphere 5 - The Foglite

it_consultant

I will give you an example. We have a production database which is located on a LUN which accessed via virtual fibre channel (NPIV in Server 2012 Hyper-V), occasionally we snap the LUN and present this LUN to another box which might be physical or virtual. We sometimes also drop the LUN off completely and present it to another host for one reason or another. You are essentially removing the virtual hypervisor from consideration visa vi your storage needs. Could be physical, could be virtual, could be Hyper-V, vmware, Xen, Red Hat...whatever.

Hyper-V Virtual Fibre Channel Overview

NPIV is not really a fledgling technology, it has been around since at least 2002.