This site may earn affiliate commissions from the links on this folio. Terms of utilize.

NAND flash

The appearance of SSDs has arguably done more to transform the feel of using a computer than whatever other event in the by viii years. Faster GPUs and CPUs benefit the high-end users that need such horsepower, merely solid state disks can exhale new performance into near any hardware. The primeval drives may have had operation issues, just once those were ironed out, it was articulate that NAND flash's ascension to the elevation of the storage market was a question of when, not if, and the "when" depended on questions of reliability and cost-per-chip — non fundamental speed. This fundamental argument has been accepted at every level, from individual PCs to high-end enterprise deployments.

That'south why it's surprising to see ZDNet instead arguing that 2022 was the "commencement of the end" for NAND flash in the enterprise. This argument hinges on a number of dissimilar papers that were published in 2022 concerning NAND reliability, performance, and suitability for datacenter applications. We covered some of these findings when the papers were new, but will summarize the collective findings here:

  • Facebook and Carnegie-Mellon found that higher temperatures tin negatively impact SSD reliability and that this correlates with college bus power consumption as well. Interestingly, this study found that failure rates did not monotonically increase (only increase) with the amount of information written to NAND wink, that sparse data layouts and dense information layouts tin both increase failure rates under certain atmospheric condition, and that SSDs that don't throttle and experience high temperatures have higher failure rates.
  • A major Korean study on VM performance plant that SSD garbage collection didn't mesh well with existing algorithms for that purpose, leading to significant functioning degradations in some cases. The newspaper concluded that it'due south currently incommunicable to guarantee a set number of IOPS when multiple VMs are hosted on a single drive. While this paper used consumer hardware, the flaws information technology found in how garbage collection is handled would have practical to enterprise equipment too.
  • A new Sandisk study found that the use of multiple layers of log-structured applications "affects sequentiality and increases write pressure to wink devices through randomization of workloads, unaligned segment sizes, and uncoordinated multi-log garbage collection. All of these effects can combine to negate the intended positive affects of using a log."

When you put these reports together, they signal to problems with SSD reliability, performance, and suitability for sure workloads. Just I'm much less sure than ZDnet that this stacks upwardly to NAND's rapid retreat from the data center.

Teething issues vs. cataclysmic deficiencies

I strongly suspect that if we could rewind the clock to the offset of the HDD era, we'd see similar comments made near the suitability of hard drives to supplant record. In the 1970s and early on 1980s, tape was the proven technology and HDDs, peculiarly HDDs in consumer systems, was the upstart newcomer. It'southward difficult to notice comparative costs (and information technology's highly segment dependent), simply the March 4, 1985 issue of Computerworld suggests that record drives were far cheaper than their HDD equivalents.

3d-nand-flash

The advent of 3D NAND flash has the potential to meliorate NAND reliability

I don't want to stretch this analogy besides far, but I think there's a lesson hither. The stride of hardware innovation is always faster than the software that follows it; you can't write software to take reward of hardware that doesn't exist yet (at to the lowest degree, non very well). It's non surprising to see that information technology'due south taken years to suss out some of the nuance of SSD apply in the enterprise, and information technology'south besides not surprising to discover that there are distinct best practices that need to be implemented in order for SSDs to perform optimally.

To cite 1 equivalent example — it was a 2005 paper (backed upward past an amusing 2009 video) that demonstrated how shouting at hard drives could literally make them stop working. While drive OEMs were plain aware of the need to dampen vibrations in enterprise deployments long before then, the result bubbled up to consumer awareness in that timeframe.

Hard drives, notwithstanding, proceed to be sold in big numbers — even in enterprise deployments.

None of this is to advise that NAND wink is foolproof or does not need a medium-to-long-term replacement. I've covered several such potential replacements only this year. Information technology does, all the same, suggest that a bit more perspective is in order. It's easy to promise huge gains on newspaper and extremely hard to evangelize those gains in a scaleable, cost-constructive fashion.

Right now, information technology looks equally though 3D NAND adoption will bulldoze the further development of SSD technology for the side by side iii-5 years. That, in turn, volition make it more hard for alternative technologies to discover footing — a replacement storage solution will demand to friction match the improving density of 3D NAND, or offering multiple orders of magnitude improve functioning in society to disrupt the NAND industry. Intel's joint Micron venture and its 3D XPoint could disrupt the status quo when it arrives adjacent year, but I'll await for benchmarks and difficult data before concluding that it will.

Far from being the beginning of the end, I suspect 2022 was the end of the outset of NAND wink, and will mark a shift towards software-level optimization and a better understanding of best practices every bit the technology moves deeper into information centers.

For more than, read How do SSDs work?