Virtual san applicance software and clustering

I have been reading about Virtual Storage Appliances (VSA) and their ability to make directly attached storage look like SAN storage.  

Here's what I would like to do with this technology :

have two servers running Windows 2008 R2

on the 'primary' server I want to have a single drive system partition running 2008 R2 and then several other directly attached drives that will serve to host HyperV VMs

on the 'secondary' server I want to have a single system partition running 2008 R2

I then want to run some type of VSA software (or any other idea that people can come up with) that will allow the directly attached drives on the primary be seen by the 'secondary' as if they were local drives as well i.e. if the OS on the primary sees 2 TB of space then the OS on the secondary will see 2 TB of space

I then want to cluster the VM's across the primary and secondary with Microsoft VM clustering

Ideally the software/technique used to turn the direct attached storage on the primary into shared 'SAN' storage will not slow access to the directly attached storage - in other words the software will be smart enough/efficient enough at the end of the day so that access to the directly attached storage on the primary server will be the same speed as if the software was not there - it would be smart enough not to put a layer of abstraction for actual directly attached disk versus non-directly attached disk.  In short I am willing to live with the fact that the secondary is slower than the primary.   The idea would be that if a VM did fail on the primary it would be ok if it came back to life  albeit slower on the secondary via VM clustering - we can live with performance degradation for a day or two.

Please note that I am familiar with VMWare and other alternatives to MS clustering/virtualization but for the purposes of this post I want to stick to the MS world of MS clustering and HyperV virtual machines and products/techniques that work with them.

Thanks in advance.