[The Register] SanDisk vows: We'll have a 16TB SSD WHOPPER by 2016

[RussianSensation]

"Intel is well behind the curve with its 10TB flash drive, slated for 2016, claims a SanDisk enterprise storage bigwig, adding that the flash memory storage provider will have a 16TB drive by then. Brian Cox, SanDisk’s senior director in marketing for enterprise storage, said the company's 4TB drive technology would develop at a 2X/year capacity pace, with an 8TB drive next year, and 16TB in 2016.

What did he think of Intel's 10TB drive news? “I think it'll be late," he said, while talking to El Reg at a recent Gartner Data Centre Summit in London."
http://www.theregister.co.uk/2014/11/25/sandisk_intel_stalls_and_well_have_16tb_ssd_by_2016/

Right now I am thinking anyone looking for a new build/OS SSD upgrade, just get a 240-512GB SSD for $100-200 and hold off until 2016 before dropping $500+ on a large SSD as it seems SSDs are about to ramp up exponentially over the next 2 years. Hopefully these advancements in enterprise SSD sizes will somewhat translate to our consumer drives.

 

[meloz]

The article is poorly written. I could not tell whether Sandisk plan to reach their goal of 16TB drive with some type of vertical nand, or write once storage.

Anyway, one thing is for sure: over the next 24 months SSDs will see the biggest improvement in GB/$ ratio yet. While SSDs will still end up rather expensive in GB/$ terms compared to HDDs in 2017-18, their performance advantage will all but kill HDDs in the data center. Those overpriced SAS "enterprise" HDDs are soon to be historical relics. Notebooks already favor SSDs due to their solid state nature. Only cheap desktops might remain the last refuge of HDDs, and for how long?

We are witnessing the death of a > US $200 billion industry, and corresponding rise of flash cartel. Samsung, IMFT, Toshiba et al will eat the hundreds of billions of 'easy money' that Seagate and Western Digital make today.

 

[RussianSensation]

Ya, I don't know what Seagate and WD are thinking. If they don't pick up their game ASAP, they will become BlackBerry/Palm of tomorrow. Sure, there will still be demand for data centers due to such high data demands, but these firms should start getting prepared for the major revolution in SSDs in the next 5 years.

 

[thecoolnessrune]

Spinners will not be replaced in the DC until the GB/$ amount reaches parity with them, which I don't see happening for a very long time. Most of the shelves in your average EMC / NetApp deployment are simply 7.2K archive drives, that aren't meant to push IOPS, but are there as the bulk storage. If we take even their most outrageous out-of-service-contract prices and go with $900 for a 3TB drive, then you're at $300 per TB. This is still far cheaper than their 800GB SSD, which costs $6,300 per drive, which comes to $7,850 per TB. Even NetApp doubled SSD capacity by sourcing HGST (who used to be STEC, which was NetApp's original SSD OEM) 1.6TB SSDs for the same price, you're looking at $3,925 per TB. If they doubled again with SSD capacity advancements, you're still looking at $1,960 per TB.

Enterprise Storage will be using spinners for a very long time.

 

[meloz]

Enterprise Storage will be using spinners for a very long time.

Reliability (durability) and performance/watt already favor SSDs. Density and cost are the only remaining hurdle, vertical nand promises to fix those. I suspect we'll see DC users jumping for SSDs long before they reach GB/$ parity with platter drives.

 

[DigDog]

there are already similarly sized SSDs and they cost $4~8k. which is completely pointless to *anyone*, because you could buy a SATA riser and plug in 8x 1Tb evos for half the price.

i imagine if there was consumer demand for this kind of flash storage size, companies would simply come out with larger SSD enclosures.

 

[thecoolnessrune]

Reliability (durability) and performance/watt already favor SSDs. Density and cost are the only remaining hurdle, vertical nand promises to fix those. I suspect we'll see DC users jumping for SSDs long before they reach GB/$ parity with platter drives.

Like I already said in my previous post, that is not going to drive a majority shift to SSDs. For one thing, we already have a vertical NAND Enterprise drive in the form of the 845DC Pro. At 800GB, the drive costs $1,000, and it costs far more in an OEM supported product (hence why NetApp drives cost over $6,000 a drive at 800GB). Assume you double your capacity or half your cost and you are still far beyond your standard spinner.

The market will continue to be based on performance necessary to get the job done at the lowest cost, which means SSD's will not entirely replace spinners in the DC for a long time. OLTP and SAP is and will continue to be moved to all-flash configurations, while VDI and bulk data storage continue to be moved to Flash Pool based systems, where a small amount of Flash storage is used in automatic tiering with a Spinner Pool. Bulk archive storage will still be on spinners.

One day flash will definitely overtake spinners, but it won't be this coming generation. As far as MTBF is concerned, that still isn't a large win for SSDs because an Enterprise solution looks at the entire pool as the solution. They don't care about any particular part. If a RAID 6 spinner pool has a high enough MTBF compared to a RAID 6 SSD when it comes to total loss of the pool (which also expands across total possible loss of a disk shelf), while costing less, then it will be chosen as the solution if the IOPS of SSD aren't needed.

 

[Hellhammer]

One day flash will definitely overtake spinners, but it won't be this coming generation.

I wouldn't be so sure. HAMR and HDMR are supposed to take HDD bit density growth back to ~30% YoY level and the current broadcast suggests 100TB hard drives by 2025. If that happens and the costs stay reasonable, flash or any other semiconductor based non-volatile memory won't stand a chance against HDDs in terms of cost. The truth is that semiconductor storage will always be relatively expensive to manufacture and thus I'm inclined to say that we won't see flash replacing HDDs as long as HDDs will continue to scale in capacity and cost.

 

[beginner99]

Ya, I don't know what Seagate and WD are thinking. If they don't pick up their game ASAP, they will become BlackBerry/Palm of tomorrow. Sure, there will still be demand for data centers due to such high data demands, but these firms should start getting prepared for the major revolution in SSDs in the next 5 years.

Or the Kodak of tomorrow. The writing is on the wall and it's huge. it's not only that hdds are slow they haven't advanced much. My pc is almost 4.5 years old and the 2 TB green drives back then cost pretty much the same as now...yes, new ones are a bit faster and there was the flood but still...almost no advancement.

 

[thecoolnessrune]

I wouldn't be so sure. HAMR and HDMR are supposed to take HDD bit density growth back to ~30% YoY level and the current broadcast suggests 100TB hard drives by 2025. If that happens and the costs stay reasonable, flash or any other semiconductor based non-volatile memory won't stand a chance against HDDs in terms of cost. The truth is that semiconductor storage will always be relatively expensive to manufacture and thus I'm inclined to say that we won't see flash replacing HDDs as long as HDDs will continue to scale in capacity and cost.

While I generally agree with what you are saying, it's not exactly cut and dry. For one thing, as you probably know, what you are suggesting is purely projection. Seagate's past CTO projected in 2005 that by 2020 we would have 40TB 2.5" HDD. Given the slowing growing of HDD densities, this isn't a likely goal. While we will definitely see a growth in density, that is not to say that the density will continue to increase at a fantastic rate. As the SSD editor, you also know that SSD technology is not sitting back and waiting

With SMR being the technology of choice until heated options become viable, this will actually set back adoption in the Enterprise market for most uses. Since the Shingled recording technology cripples write speeds, they'll be mostly adopted in Backup / Archival areas (like through Avamar products and similar). But in the middle ground (VDI), it will only push SSD-based tiering further as densities continue to increase, while storage drive density will remain the same (I expect Enterprise markets to move towards 4TB and eventually 6TB drives, but they'll stay there as long as higher densities require SMR). One we get HAMR and HDMR though, I agree that spinners will get a massive boost in densities, which will be very welcomed by the market.

As increasing densities lower IOPS provided by HDDs (due to fewer drives needed in a standard pool), flash caching will be relied upon even more to keep that balance between storage space, and storage input/output.

In the long interim, the future is bright for both technologies! As stated before, neither of them are going anytime soon

 

[VirtualLarry]

With all of that said, will we see more tiered-caching support in the prosumer DIY server market? FreeNAS/ZFS, unRAID, etc?

 

[wli]

From 1 to 16TB in two years, I will be a very happy consumer to upgrade.

 

[StinkyPinky]

It won't be long before HDD's are obsolete imo. SSD's are already dropping quite a bit in price and their performance just blows HDD's away. THere will come a tipping point where OEM's like Dell and Lenevo etc decide that putting in SSD's into even budget computers is feasible and once that happens the market for HDD's will start to tank.

I'm giving it three years for that to happen

 

[mikeymikec]

On one hand I can't believe that companies like Seagate and WD aren't prepared for what comes after HDDs in the mainstream, yet on the other hand I think of how badly they screwed up their planning, which showed itself perfectly well with the flooding in Thailand. Honestly? They had a single linchpin to their operations?