Sorry, one more post. We're looking at getting something like this for the office:
https://www.qnap.com/solution/thunderbolt-nas/en-us/
Thanks for this. It's an interesting capability that's a bit different (& better in some ways) than a Promise Pegasus style cabinet ... although it too illustrates my point about the value paradigm erosion of the nMP versus its predecessor.
Case in point, as per a quick read of B&H, it looks like a bare (no drives) copy of this 8-bay NAS runs $3200, which on an (overly)simplistic view means that the lost four (4) internal bays of the cMP were a consumer value worth $1600.
------
FWIW, a side comment on several other posts' comparisons, for which I'm going to use the above comparison's hardware mismatch as an illustration:
Part of the reason why I dentified the above as an over simplistic comparison is because it looks like that NAS is using 10Gbit Ethernet, whereas both MP's are single GBbit Ethernet. True, but let's also not forget that technology ... and the cost thereof ... marches on: Apple is pretty much overdue to bump up from Gb to 10Gbit for awhile - especially on highest end gear. And this is my allusion to other post comments: we should recognize that a notional cMP "do-over" (let's call it a '2016 cMP') would use the appropriate technologies of the day and not let it be stuck with what the 2012 shipped with. This is where 'cMP with TB' statements are coming from, plus let's also not neglect history either: the 2012 cMP was a strongly panned 'update' for all of the contemporarily-available updates that it lacked - two big examples noted st the time were the lack of updates ...back in 2012... to SATA-3 and USB3...because it literally wasn't a new machine, but merely a CPU plug-in on the old motherboard.
To that end, any notional exercise of a 'do-over 2016 cMP' would certainly do better than be still using that old ~2009 vintage motherboard. For what this could entail is a reasonable debate, but the long & short is that even stuff like TB isn't out of the question, because Apple was a development partner, so if they really wanted it to happen, it would --- just like SATA3 and USB3, its configuration constraints is a business policy barrier, not a physics one: just like the decisions for which ports to delete from the MacBook, which video adapters to not design/sell, etc....
-hh