However, I'm not sure making the Mac Pro rackable will help all that much. It certainly would for some, but after Xserve's death, I don't see too many people using OSX in a cluster-type set up.
The XServe was dead in cluster-type set ups when it fell behind high speed interconnect (e.g., Infiniband) long before it was discontinued. That whole drive into a "supercomputer cluster" land wasn't particularly well thought out strategy.
This doesn't necessarily have to do with a cluster. There are numerous usages for a Mac Pro inside of just a 1/4 , 1/2 , or whole rack that are oriented around usage of OS X Server on jobs a Mac mini can't handle. The problem now is 1/4 and 1/2 rack set-ups are impeded by the vertical orientation now.
This issue is that a single ( or dual with fail-over ) server set up would work better in a rack. Place the Mac Pro (or two ) in a rack with some DAS (or NAS/SAN) storage units and perhaps a tape unit and you have a SMB IT infrastructure in a single rack. The Mac Pro could be sliced up with a hypervisor to offer up the 3-6 servers a modest sized shop would need. There is zero need to drift off into the "large enterprise" and/or multiple "cloud rack" kind of solutions.
Putting OS X Server aside, the Mac Pro may be relatively quiet for a workstation of its caliber it may need to be attached to a "banging and clanging" storage or I/O box in a rack in another room. Nor is the vertical set up conducive if have a semi-mobile 1/4 height set-up that move from location to location.
Finally, if a shop needs just one or two software update servers and/or a "mega large" Time Capsule storage server then even just a singular Mac Pro horizontally racked up with the other myriad of Linux/Windows boxes is a better fit.
The HP workstations are rack mountable. For the most part the Dell ones are also. It isn't like other vendors haven't already implemented the features and they aren't being leveraged at least as much as folks significantly leverage both pairs of the MP handles.
A Mac Pro with OSX works great in combination with a linux based cluster. And the linux cluster gives you a lot more power for the money once you get above the $7000 workstation price range.
This isn't all about purely computational benchmarks. In that context the Mac Pro looses traction. If batch rendering jobs are primary bottleneck then yes those will drift off the Mac Pro. There is really no huge need for a purely local GUI or purely local data storage.
Collapsing workload onto fewer boxes has been a large and growing market (e.g., virtualization ) as much as "scale out" clusters has been over the last 4-5 years. Frankly, this is one of the systemic problems of the Mac Pro; overly narrow concepts of where it can be deployed effectively.
So if it was a relatively cheap change for Apple, yeah do it. But I'm not sure if up rooting the current production line would be worth it.
In between motherboard revisions, sure it would make no sense. However, the new E5's require a new socket and hence new board. The internals would be shifting anyway. The case is just how the internals interface with the outside world. There are other dynamics going on ( ODD out of favor , SSDs , Thunderbolt , USB 3.0 , GPGPU upswing, etc. ) all of which allow consolidated usage of a Mac Pro coupled to other boxes in a rack to tackle specialized missions.
).
