Seagate, WD need to get a firm grip on solid disks
Pro-tip: They should buy OCZ and STEC
Blocks and Files STEC is touting its great CellCare flash endurance, and it is great, reinforcing a view that it and OCZ make natural flash market buy-in points for Seagate and Western Digital.
Whoa, hang on a minute, where did that idea come from?
Let's start from the point that the key future flash technology is the ability to lengthen the endurance of MLC and coming TLC (3-bits per cell) flash.
I'm told OCZ's Ndurance technology can make unusable raw TLC into usable TLC product.
STEC has just announced its CellCare technology can extend raw 24nm MLC life 13.3 times, from the start point of 3,000 program-erase cycles to 40,000 P/E cycles. It believes its 400GB MLC SSDs, due later this year, can sustain ten full capacity writes a day for five years with no performance degradation. Imagine that technology applied to TLC flash, a 13X uplift on the raw P/E cycle maximum, and you will be able to get usable TLC solid-state disks from STEC.
Startup DensBits says it can make TLC flash last for 10,000 write cycles (P/E cycles actually), but it is a very new startup and has no shipping product.
Half-hearted HDD suppliers
Moving the argument on, let's say that, in general, Seagate and WD are half-hearted about SSDs. Their business is making spinning disks to store bits and they don't want to cannibalise disk sales with SSD sales. They are, as it were, reluctant SSD sellers. We'll admit WD's Hitachi GST subsidiary is furthest ahead with its Fibre Channel-connected SSDs developed using Intel technology - but WD/HGST needs to cut its Intel apron strings as both companies compete in the SSD business.
Further, we'll say that neither Seagate nor WD have leading flash controller technology.
STEC developed its own and got to be EMC's and IBM's SSD flavour of the month for a couple of thrilling years. Apple recently picked up Anobit, SanDisk picked up Pliant, LSI picked up Sandforce, and OCZ picked up IndiLinx.
No supplier can make a great SSD or PCIe flash drive without owning their own controller tech and closely integrated NAND chips. Agree with that thought and both Seagate and WD need flash controller tech that can give MLC and the coming TLC flash drives a reliable, long and dependable working life.
Now I'll bring in another thought; successful flash drive sales need a working OEM and sales channel. Both STEC and OCZ have these, with STEC already in the enterprise and OCZ coming at the enterprise from its existing consumer operation. It's thought, for example, that OCZ may be supplying some flash array vendors with flash drives, and could be selling Everest controllers to third-parties. OCZ may also be in the Open Compute 2 initiative.
Another assertion now; looking long-term the proportion of online data held on flash as opposed to disk will rise. TLC flash will be added to MLC and SLC (single-level cell) and a post-NAND technology, such as Phase-Change Memory (PCM) or Resistive RAM (RE-RAM) will continue the on-volatile memory charge when NAND runs out of steam at sub-10nm process geometry levels. The type of data held on flash will develop.
Currently hot primary data is moving to flash. Eventually all data that is not nearline will move to flash. The disk-to-flash migration over time, measured in decades, will parallel the tape-to-disk migration. An extreme view is that eventually nearline data too will move to solid state.
Bit storage colonies
So unless the disk drive manufacturers redefine their business they will not be successful players in the non-volatile solid-state storage business. They have to establish colonies in this business now.
I understand from sources, not from Seagate or WD I hasten to add, that both Seagate and WD are thinking of buying their way more deeply into the flash drive business. Two good targets for them are OCZ and STEC. Violin Memory is tied to its own flash drive technology and Fusion-io, which most people assume will be bought, is deeply embedded in its server flash business, not the wider solid state storage business; so I'd rule them both out of contention from the Seagate and WD point of view.
Looking at cultural fits we enter the match-making business. Would Steve Luczo's exec team at Seagate get on better with the Moshayedis at STEC or Ryan Petersen's team at OCZ? Which of the STEC and OCZ teams would WD's John Coyne-led execs deal better with? One view is that Luczo and Petersen are a pair of "very" smart cowboys while WD's Coyne is a canny, canny older guy, leading to Seagate-OCZ and WD-STEC matches.
A wildcard is Micron. It may well like the idea of owning its own controller technology. Any such Micron ambitions may be put on the back burner now because of its current involvement with bidding for Elpida. One other wildcard is Toshiba which may well also like the idea of having mainstream and advanced flash drive controller technology it could integrate with its own chips.
One thrust of this argument is that vertical integration is coming in the flash business and the HDD manufacturers, the three surviving drive manufacturers know all about that and all about what happens to suppliers in a consolidating market that miss the boat.
That's the case; Seagate and WD need to get into the flash controller market. That technology is moving to extracting every last bit of useful life out of MLC and TLC flash, and OCZ and STEC are the only companies with shipping controllers, existing OEM and other channels, and MLC and TLC endurance-maximising technologies. Ergo quod erat demonstrandum. ®
Comments to this forum topic, please. ®
Sponsored: Optimizing the hybrid cloud