Thanks for the suggestions Jase. I think in our case I would like to stick with a linux option. I have ran 2k3 r2 and 2008 as an NFS target and have not been real impressed with the stability. Whether performance that seems to fluctuate for no easily traced reason or just dropping connections altogether to the point of having to restart to the NFS Server services. I have not experienced this problem with Redhat and we have someone very intimate with the product that can performance tweak for us.

My biggest concern is how well a solution like this could scale with the appropriate disk subsystem in place. I was thinking an 8-10 disk raid 10 array with 15k sas drives whether that is in a pe2900, or attached by an md1000 or pv220. It would be with ample processor, memory and NIC's, likely two teamed at a 1gbps a piece so i would image throughput shouldn't be a huge issue?!

I hear ya. And it sucks. Coming from an environment with a larger investment in technology to an environment 3 x times the size, server wise but lacking any kind of centralized storage is kind of mind boggling. We were running an Equallogic PS300e and CX3-10 at my last employer and here...well EVERYTHING is in-line or direct attached storage for 350+ servers