|
|
|
|
|
|
|
|
|
|
xen-devel
RE: [Xen-devel] NPIV support
> I've suddenly found myself in a position where NPIV support in Xen would
> be desirable to say the least. I notice several distros have included
> NPIV support, I'm assuming through their own extensions as I can't find
> examples in my distro nor in the Xen 3.4 source/examples. It seems
> strange the other distro's changes haven't made it to the official source.
> I suppose it does make sense if the 2.6.18 kernel doesn't support npiv.
>
> I'm wondering if this is a hot topic on the development side of things, or
> if it's been passed over for lack of developer time/resources. If the
> latter, I'm sorely tempted to take a stab at it myself if you could point
> me in the right direction. Or is this something that is waiting for Dom0
> inclusion in a future linux kernel release?
I'm not sure which kernels first started supporting NPIV, but you'll certainly
need to be on something newer than 2.6.18. I'd look to the XCI 2.6.27 tree.
Adding NPIV support should be fairly straightforward, particularly if you add
it to the PVSCSI code which already has a notion of host adaptors exposed to
guests. Just add a way of assigning a WWN to a virtual HBA, and then add
support in the backend to use the NPIV sysfs interface to instantiate a new
WWN, and then export all the LUNs visible by that WWN to the guest's virtual
HBA.
It would also be possible to do something similar with the blk protocol, which
would likely be fine if your intention was just to expose disk LUNs, and you're
just using NPIV for zoning and QoS differentiation benefits. This is likely to
perform rather better as the PVSCSI protocol hasn't had any of the optimization
that the PV block protocol has.
Ian
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
|
|
|
|
|