Marketing baby!
Since when was marketing reality?
EMC, we have a problem. Your Project Nile has exabyte-plus capacity and is built from EMC's ViPR control/data plane software and VNX arrays. Yet, the biggest VNX is the 8000 at 4.5PB capacity, about 223 times too small. What gives? EMC COO and president David Goulden said in Milan that Project Nile will use ViPR and VNX to …
Nile is a pre-announcement. It's two years out. In that time, expect 10TB drives and perhaps more. Seagate is working on shingle and HAMR, and may have "2D" storage too. WD is working on technologies as well. Helium-filled drives may have 50% more platters, and they are looking at shingle and HAMR.
Furthermore, the Nile architecture will involve a lot of head nodes on those VNX arrays. I'm going to predict that exabyte-class systems are almost all Ethernet-accessed Object storage, with localized zones of BlockIO. They won't be jumbo SANs.
If helium allows platters to be closer together, it is likely they'll just make the drives thinner while keeping the number of platters constant. That is, after all, what they've always done in the past when the ability to put the platters closer together has presented itself.
If you can fit 50% more drives in a rack, the outcome is the same, of course.
Every EMC sales guy I have ever had the fortune of speaking to (and they knock on my door weekly) always talks values post dedupe. If you go and look at the specs for the datadomain devices for instance:
DD2500
Up to 6.6 PB Logical Capacity
Up to 133 TB Usable Capacity
It only give you 133TB RAW, but the sales guys can tout a 6.6PB figure. Apply that scale of expected dedupe to the original problem and your 233 racks becomes 12.