a winter's tale of SSD market influences - from industrial flash controllers to HPC flash arrays

thought leadership in the SSD market

from - what's in the chips?

to - how will SSDs change the future?

Since the 1990s our readers have been accelerating the growth of the SSD market and setting its direction and agenda.

.....

If you could go back in time and take with you a factory full of modern memory chips and SSDs (along with backwards compatible adapters) what real impact would that have?

are we ready for infinitely faster RAM?

.....

SSD news
SSD jargon
the fastest SSDs
2.5" NVMe SSDs
controllers for SSDs
the top SSD companies
endurance myths legends
history of the SSD market
will SSDs end bottlenecks?
can you trust SSD market data?
DWPD - examples from the market
surviving SSD power loss - all SSDs
branding strategies in the SSD market
capacitor hold up extremes in 2.5" SSDs
trends in the enterprise PCIe SSD market
enterprise adoption of flash - 2004 to 2016
spinning down to HDD's market retirement
lost markets for enterprise rackmount flash

what were the big SSD ideas which emerged in 2016?

"Storage Class Memory

As storage class memories are emerging, the memory hierarchy will be changed. NOR-based NVDIMMs, such as 3D Super-NOR and 3D XPoint, will replace DRAM and SSD at the same time.

Also, software-based NVDIMM-P, such as HybriDIMM, will come to the storage class memory market. Storage class memories mingles fast-but-expensive volatile, and slow-but-inexpensive non-volatile memories together. As a result, it will significantly boost system performance at low cost and create huge market opportunities."

Sang-Yun Lee, President & CEO - BeSang

...read more in the article - big SSD ideas which emerged in 2016

image shows Megabyte reading a scroll - the top 50 SSD articles and recommended SSD blogs

high availability SSDs


SSD endurance


leading the way to the new storage frontier	.....


SSD news	..

a winter's tale of SSD market influences - from industrial flash controllers to HPC flash arrays - set against the tapestry of a single company which divided to face the tempests of change
by Zsolt Kerekes, editor - StorageSearch.com - November 14, 2016

A conversation I had last week with Charles Tsai, President - AccelStor provides a useful example of how, nowadays, even the simplest type of SSD product plan has to be aware of strategic considerations in a wide range of contexts which are sweeping across the whole market. The factors which shaped everything in our conversation being:-

strategic changes in the merchant SSD controller market.

Competing pulls between greater standardization, customization and for many companies the need to exit the controller market altogether and offer solutions at higher levels of integration.

strategic changes in the enterprise systems market.

When too many bundled value-add features are perceived by users as fool's-gold which merely add cost and subtract value.

strategic changes in the SSD-aware software market.

Rewriting every assumption which once touched a hard drive.

emerging changes which will come from the DIMM wars and SCM markets.

If SDS users can eat NVDIMMs than flash array makers also need to look at eating such dog food.

The last time I spoke to Charles had been 2 years before (June 2014) - when I had been curious to find out what his company InnoDisk (an industrial SSD maker) thought it was doing by introducing a rackmount SSD (FlexiArray) for the high performance enterprise market.

I wrote up some of that stuff in my (2014) article - decloaking hidden and missing segments in the analysis of market opportunities for enterprise rackmount flash. InnoDisk spun out that product line and its associated marketing as a separate business called AccelStor in November 2014. And in August 2016 - the company's FlexiRemap software technology (which is used in its rackmount SSDs) won the Best-of-Show Technology Innovation Award at the Flash Memory Summit.

So the first thing I said to Charles was that he had been right about a lot of things he had said in our earlier conversation and that the launch of a new enterprise SSD box business by a company which was previously known only for its SSD controllers and industrial SSDs - a move which seemed rash or risky at the time - now in hindsight made more sense as fitting into the kind of framework which all SSD companies have to be aware of.

In the context of strategic changes in the merchant SSD controller market - all companies face the challenge that there is only so far that they can go in getting mileage out of innovative features in standard SSDs.

Most controller companies will have to make it easier for drive makers to customize their products and to live with the idea that - in an array setting - it is only possible to get best utilization of the flash from a system level aware perspective.

Trying to put too many clever features into a solo SSD can indeed be counter-productive because designers who integrate SSDs into arrays can get better results in other ways.

And in fact Pure Storage has said- in a 2015 paper (pdf) on HA design using flash - that when evaluating commodity SSDs for use in its own arrays that the introduction of clever benchmark driven controller activity which makes an SSD perform better at the solo (consumer market) level means that the SSD performs worse at the array level than simpler controller variants.

The business outlook for merchant controller companies is that business won't be the same as usual.

And confirming the trend which was one of the big SSD lessons of 2013 - I can reiterate that the box (or array) level of integration is still where SSD technology cleverness can be best experienced, judged and valued at a business level. (Another way of saying this is that SSD drive makers are hostages to the fortunes of the arrays their drives populate. That's why enterprise drive makers have been entering the rackmount systems market to participate directly. Meanwhile for many other drive makers - the array - at cloud level - represents a volume business opportunity in which being a well behaved member of the array population is a key citizenship criterion.)

Here are some of the things I learned about AccelStor's business and products in my recent conversation with Charles Tsai.

The separation of InnoDisk (focused on industrial markets) and AccelStor (focused on enterprise markets) into 2 distinct businesses although sharing core flash memory expertise enables each business to follow marketing plans which are appropriate for each market.

AccelStor's products are optimized for performance (as opposed to capacity).

Charles said the high performance flash array market is growing fast with many new customers and new types of applications.

Most of the efforts to improve the product line in the past year or so have been directed towards improving the internal software. Each flash array system is inside a micro-tiered system. (In the past when we encountered the term "micro-tiering" in a storage box or SDS - this was often in the context of a hybrid storage array. But as I explained in my 2012 article - 7 ways to classify where all SSDs will fit in the pure SSD datacenter - there will be more (cost based) latency zones in the pure solid state storage and memory world than existed before in the rotating storage era - hence - even more need for tiering, caching and bridging technologies.

High availability in SSD systems is architected in different ways by different vendors. AccelStor has been drawn towards implementing HA with the fastest possible time to continue operations with failed system. So they use 2 node active-active synchronous replication with a shared nothing design. The result is that after the failure of an entire node then data IO can continue on the operating node in under a microsecond.

From a traditional storage point of view such a failover scheme can appear to be expensive - but Charles said this is what AccelStor's customers prefer.

I said to Charles I thought that the old maths of high availability costs were becoming more complex now because users could gain virtual capacity by leveraging fast flash storage speed (in their software) and might decide they can afford to trade off such gains against the cost of having more hardware.

Some applications can tolerate longer recovery times than others. It's one of the many areas in which customers have different needs and will self select different approaches.

When I heard about the many levels of micro-tiering in AccelStor's arrays I asked what the company was doing about NVDIMMs etc.

Charles said that when it came to the internal design of the NeoSapphire flash arrays the optimization for performance didn't start and end with the flash controllers. He said that the company's knowledge of memory and controller design reached into every part of the box and that included looking at the roles of flash and DRAM and where non volatility could be beneficial and leveraged in software.

This is what I expected - given what Charles had said in an earlier part of this conversation. And also because nvRAM techniques have been used in flash arrays and software defined storage systems by other vendors in the past.

The issue of innovation arose several times in my conversation with Charles Tsai. In effect the message I took away was that unlike traditional storage array companies which have to fight hard to retain even a fraction of their previous revenue (due to theincreased utilization and effectiveness of solid state storage and SSD aware software - which means users can do more with less) newer flash systems vendors such as AccelStor can instead identify which markets are growing with new customer needs and design products for those markets without worrying about the impact on older product lines.

Charles also said that it's necessary for companies to compete with what they did before - because if you don't innovate then someone else will. But that's easy to say when you don't have a big vested market position.

I said in our conversation that some vendors like SanDisk faced the dilemma that by doing what was right for the future of the whole business (such as introducing rack based flash systems) they lost business from some traditional buyers of their enterprise SSDs who saw them as a potential competitor. But in a market like SSDs where so many clever technology companies are competing you have to anticipate what will be in the best interests of the customer and go with those product decisions regardless of the short term consequences (as long as your company can ride through to the next technology generation).

In some ways that's a business gaming strategy similar to the DRAM market. But unlike DRAM where huge investments of resources are necessary to claim a place at the gaming table - the SSD market still has dice which are loaded in favor of startups (until their success spawns a new generation of imitators and replacements).

How long can this game go on?

The remarkable thing is that from some perspectives everything in the SSD market still looks like it's ready to change while from other views it seems that nearly every technology decision has already been settled. (See the SSD heresies for exampes.)

The onset of a predictable stable future for solid state storage is what makes the rush to change everything before it settles down so urgent and rewarding.

In a way what my conversation with AccelStor shows is that every aspect of an SSD product:- from what it should do to how the functions are implemented internally - is affected by competitive and collaborative ecosystems influences in other markets.

Guessing the future for the whole market makes it easier for vendors to identify what their best fit roles might be. And as we've seen in SSD history - adaptation requires big changes. Such technology changes have to be attuned to needs that customers recognize and are willing to buy - before the funding for new product lines burn out.

One of the signs of growing diversity in the SSD ecosystem is the growing willingness of some vendors to ignore merchant market research predictions (which are often cast in terms which are too wide or backwards looking to be useful) and instead for vendors to develop their own understanding of micro segments based on customer feedback.

Big versus Small in SSD controller architecture (2011) - In any SSD system (whether it's a drive, array or cloud) the limits of what is possible to achieve in reliability, performance and efficiency is dictated by the underlying controller sizing philosophy.

Drivers, mechanisms and routes towards consolidation in the enterprise SSD market (2015) - Users will increasingly interpret all data systems as fitting into a spectrum of commodity like boxes with recognizable colors of latency, capacity or bridging capabilities. Consequently they will gravitate towards a reduced set of system types and software.

Where are we heading with memory intensive systems and software? (2016) - When you start taking polymorphous memory systems architecture seriously as a permanent fixture in the data systems landscape then the future of computing will look very different to the past. And there will be greater expectations about what SSD systems can do.

Hmm... it looks like you're seriously interested in SSDs. So please visit our home page and come back again soon.

....

It was the best of times.
It was the worst of times.
2017 was a year like nothing before in SSD history.

which way next for SSD?

Some suppliers will quote you higher DWPD even if nothing changes in the BOM.

what's the state of DWPD?

Is more always better?

The ups and downs of capacitor hold up in 2.5" flash SSDs

Retiring and retiering enterprise DRAM was one of the big SSD ideas which took hold in the market in 2015.

Over 20 companies have already announced products for this market among which are Memory1, 3DXPoint etc

But what are the underlying reasons that will make it feasible for slower cheaper memory to replace most of the future DRAM market without applications noticing?

latency loving reasons for fading out DRAM

....

Software is the data heat pump which can transform the entropy of flash memory arrays from mere storage systems into higher value random access memory.

Where are we heading with memory systems and software?

....

In SSD land - rules are made to be broken.

7 tips to survive and thrive in enterprise SSD

is data remanence in NVDIMMs a new risk factor?

maybe the risk was already there with DRAM

There's a genuine characterization problem for the SCM industry which is:- what are the most useful metrics to judge tiered memory systems by?

is it realistic to talk about memory IOPS?

Many of the important and sometimes mysterious behavioral aspects of SSDs which predetermine their application limitations and usable market roles can only be understood when you look at how well the designer has dealt with managing the symmetries and asymmetries which are implicit in the underlying technologies which are contained within the SSD.

how fast can your SSD run backwards?

....

The enterprise SSD story...

why's the plot so complicated?

and was there ever a missed opportunity in the past to simplify it?

the elusive golden age of enterprise SSDs

....

How committed (really) are these companies
to the military SSD business?

a not so simple list of military SSD companies

Why do SSD revenue forecasts by enterprise vendors so often fail to anticipate crashes in demand from their existing customers?

meet Ken and the enterprise SSD software event horizon

Reliability is an important factor in many applications which use SSDs. But can you trust an SSD brand just because it claims to be reliable in its ads?

the cultivation and nurturing of "reliability"
in a 2.5" SSD brand

....

unraveling the unreal positioning of AFA startups

compared to EMC...

Now we're seeing new trends in pricing flash arrays which don't even pretend that you can analyze and predict the benefits using technical models.

Exiting the Astrological Age of Enterprise SSD Pricing

It's by no means inevitable that the biggest memory companies will go on to become the biggest SSD companies.

an SSD guide to semiconductor memory boom-bust cycles

Adaptive dynamic refresh to improve ECC and power consumption, tiered memory latencies and some other ideas.

Are you ready to rethink RAM?

90% of the enterprise SSD companies which you know have no good reasons to survive.

market consolidation - why? how? when?

We can't afford NOT to be in the SSD market...

Hostage to the fortunes of SSD

Why buy SSDs?

6 user value propositions for buying SSDs

"Play it again Sam - as time goes by..."

the Problem with Write IOPS - in flash SSDs

Why can't SSD's true believers agree on a single shared vision for the future of solid state storage?

the SSD Heresies

The memory chip count ceiling around which the SSD controller IP is optimized - predetermines the efficiency of achieving system-wide goals like cost, performance and reliability.

size matters in SSD controller architecture

Aeon, Alacritech, ArmourDrive... etc

Inanimate Power, Speed and Strength Metaphors in SSD brands

A popular fad in selling flash SSDs is life assurance and health care claims as in - my flash SSD controller care scheme is 100x better (than all the rest).

razzle dazzling flash SSD cell care

If you've seen or read - The Hobbit - then you'll be familiar with the concept of the riddle game. Something similar is playing out now in the enterprise flash array market.

playing the enterprise SSD box riddle game

a winter's tale of SSD market influences - from industrial flash controllers to HPC flash arrays - set against the tapestry of a single company which divided to face the tempests of change