E.g. rather than have a mess of external drives scattered over your desk, someone can make a box with 2-4 bays to take a combination of HD, SSD and optical drives.
How many Mac Pro users have a mess of external drives scattered over their desk now? Relatively few.
You create a problem and then introduce Thunderbolt to solve the problem just created. That is not a good-fit application of technology.
I didn't say that - but all that really needs to be in the main box is the system drive - which is increasingly likely to be a SSD.
The unsupported flaw here is the presumption of this being the requirement ("need").
If 40+% of current Mac Pro users are using more than 3 drives that isn't really supported by what users are actually doing. Sure, if 70% current Mac Pros had 2 or less volumes mounted ( i.e. most folks were running RAID-0 and maybe another drive ) then sure a single SSD probably solves that issue. In that case, most users were previously short-stroking 2-3 drives with RAID-0 ( or RAID-5) striping to get around rotational latency of the HDDs. SSDs largely solve that and the capacities have risen to the point if short-stroking ("throwing away capacity") of a HDD the capacities largely now match ( with a tractable price premium for the SSD solution).
However, if the majority of Mac Pro users have a data size needs (i.e., TBs of space) then the single SSD does not solve the problem (and won't cost effective do so for several years ). You would have added a distinct non-competitive limitation to the system. Systems that more cost effectively allow for 8-12TB capacities will win in product selection competitions.
Users use completed, deployed systems. The requirements are best matched by looking at what the deployed systems are likely to be over a significant sample size. They are not accurately driven by looking a small, narrow samplings.
For some users, that's all you need. For other users, you attach an external enclosure tailored to your need (HD/SSD, speed, redundancy, interface...)
It can't be just "some" users. It has to be "most" users. The
standard features of a system must be a value-add features for most of the potential users for the system in general to present value to the potential buyers.
...mainly for high-end graphics cards. I did say they were the fly in the ointment. If they can find a way of fitting those externally then the "base" box can be smaller, lighter and quieter.
Bandwidth is not just high-end graphics cards. Mac Pros hooked to a modern SAN/NAS network that is 10+Gb/s with redundancy are also high-bandwidth consumers. The external box doesn't have to be sitting on the desktop. More than a few Mac Pros are hooked to external boxes that aren't a simple Thunderbolt cable distance away.
Personally, I don't need a full-blown workstation-class machine but a "super Mini" with similar specs to a high-end iMac would be interesting -
Ah the "backdoor" xMac surfaces yet again. Apple composing a xMac will make them largely uncompetitive in the workstation class space. A headless iMac doesn't have anywhere near the bandwidth or the "horsepower" of a workstation class machine. Slapping Thunderbolt on it and tap-dancing claims that addresses the issue are just smoke. It does not solve the missing bandwidth or "horsepower" problem at all. Even the upcoming Falcon ridge update in 2014 won't.
There is a gap in the $1899-2,499 zone that Apple has right now for another "box with slots". I think would be a mistake to mutate the Mac Pro to primarily fit that space. That space probably needs a different product. Whether they 'merge' that product into the Mac Pro umbrella space ( e.g., into "Mac Pro S (shorter)" and "Mac Pro T (taller.. i.e., classic)" ) or just use a different name is up to them.
A Xeon E3 box with Thunderbolt (integrated graphics) , 10GbE sockets , and 1 x16 ( or perhaps two x8 sockets electrical ) would be a better match for those who don't like the top end BTO iMacs but don't quite have needs for a Mac Pro.
