what kind of SSD world... 2015?by
editor - StorageSearch.com
- January 2015
|SSD year 2015 in the
enterprise already offers the tantalizing promise that even on those rare
occasions when most people can agree they will be talking about the one same
thing - you can be sure by what follows that everyone had a completely
different idea of what that same thing meant to them. |
defined storage) being a good example of an illusory market idea for
comfortably bringing technologies together which due to its many incompatible
incarnations will probably rip many business plans apart.
market has been around for a long time -
years since the start of the ancient era and
12 years since the
modern era of SSDs - and there is a tendency for all tech markets to slow
down and become more predictable.
Following that line of thinking many
readers - yearning for relief from their SSD uncertainty induced headaches
- have asked me if - as a result of what they called recent "consolidation"
in the market (by which they meant
what did I think about the prospects of any stability and order coming into
And did I anticipate any drastic reduction in the
number of SSD companies and new SSD startups?
My answers were:-
change is coming
- no - I don't think we've reached stability in reference enterprise SSD
designs and use cases.
All the systems in the market today are
implementations of transient architectures which are pragmatically adapting
legacy installations and infrastructure.
Let me clarify that when I use
the word "transient" to describe different identifiable phases of the
SSD market - what I mean is time periods which are typically less than 5 years
and always much less than 10.
It has always been clear to me that
it would be impossible for the SSD market to get from its original stable
embryonic start (tiny niche market) to the final stable state (SSDs
everywhere as the dominant data technology by 2020) without going through
multiple transient phases along the way.
That's because the
investments needed to solve known technology problems only make economic sense
when there is already a big enough ecosystem to support the cost.
each transient phase of the SSD market requires a different bunch of
competitive challenges to be met - not least of which is education.
there is the ever present risk that some roads - which look attractive along
the way (and which generate the kind of revenue spikes which spawn crazy
IPOs) - will prove to be dead ends.
Having said that - it's also self
evident that even when we do get to the state when SSDs represent the biggest
part of users' new hardware budgets (and we aren't there yet) there will still
be an economic case for adapting older SSD systems - into new setups.
Some types of products in the market today will make this
transition better than others.
The obvious examples being server based
systems (due to their inherent flexibility) and whatever SSD storage products
become the de-facto standards (due to their widespread user deployment).
You probably won't be surprised to hear me say I
think there are still many surprises ahead from SSD market in 2015 - which
promises to be another momentous year for the SSD market.
As long ago as 2010 - it was clear that the
last big user adoption hurdle which enterprise SSDs hadn't solved yet was their
use as low cost alternatives to arrays of hard drives.
My gut feel was
it would happen at the system level - despite all the extrapolated media
trends at that time pointing towards the fact that hard drives would
continue to offer lower cost per raw terabyte at the component level
compared to flash SSDs.
Starting with the
condition assumption that HDD drives would have zero purchase cost per
terabyte - I explained in my classic article -
this way to the
petabyte SSD - why the deciding factors would be a combination of
operating costs (smaller physical space, lower power consumption) and the data
What do I mean by data utility?
How useful is the
product from the data factory point of view. And when comparing SSD based
storage systems to HDD based arrays - the particular significance of "virtual
usable capacity" outweighing raw physical capacity - due to the fact that
a raw terabyte of SSD (with efficient dedupe and compression)
can support more
users than a raw terabyte of magnetic storage - even when running the same
In that 2010 petabyte SSD article I stuck my neck out
and predicted that 2016 would be a significant crossover year for the storage
industry - characterized by an iconic new type of product - the availability of
a petabyte of SSD utility storage in a 2U rack.
With the benefit of
hindsight I was too pessimistic about how well controller designers would manage
flash memory - and in 2014 I revised my estimate for that crossover year as
being sooner - in 2015.
Although there will be many other
revenue shifts in the storage market in 2015 due to the mainstream adoption of
enterprise SSDs following predictable patterns and trends - I think that
assaulting what I called back then - the last bastion of the enterprise hard
drive - will still rank as being the most important of the many long ago
predicted market milestones for 2015.
the unpredictable stuff
predictable in detail - will be SSD user reactions to new dimensions in
The "one legacy server/storage
architecture" to fit all "general purpose applications"
enterprise model is clearly an economic and technical
illusion as more SSDs
enter the picture.
What is less clear - however - is the degree to
which SSD based offerings - which simply ape and mimic
architectures will make up the final picture.
Making up the current
market mix there are products which interoperate and span HDD and SSD
generations of the installed base - while other new products promise to be
better replacements for specific applications.
Adding to the
confusion of what is possible in the SSDserver mix while retaining old
applications software and classical boundaries between servers and storage will
the growing market success of entirely different vendor interpretations of SDS
(software defined storage) and fabric spanning server technologies which
leverage SSD latencies and at heart include new generations of thinking about
I think "SDS" is an almost useless top level label for many of the
products which fly under the SDS flag.
That's because there are so
many different solution which are being offered under the SDS colors which are
very different in what they do and how they do it.
What we really need
is more jargon and
granularity of definition which will help to categorize SDS products into
similar and dissimilar buckets.
This isn't a new problem for the SSD
10 years ago - a simple bilateral split of SSD types by
memory type - DRAM or flash - was a good enough clue to indicate almost
everything you would need to know in advance about a new SSD and the markets
for which it might be intended.
But it's been years since such
headline flash descriptors told you anything definitive about the market use of
The list of things which you need to safely
understand any new
SSD drive (even a humble old style embedded legacy CF or USB drive) let alone
system has grown very long indeed.
So don't be surprised to see
the SDS market growing up this year and accruing many more sub levels of
descriptors and architectural shortcut descriptors. Because under the guise of
new "software" - what we're really getting is glimpses of new
SSDcentric visions of what is possible with SSD enhanced software architecture.
a story which will retain our rapt attention well into the next decade.
the meantime - here in early 2015 my firm belief is that the SSD market
will demonstrate that it still retains the ability to surprise and shock us
all. (Myself included.)
SSD market volatility in 2015?
a future article I'll look at whether it's possible to predict the comparative
volatility of the SSD market in 2015 - based on the empirical measure of how
many new companies have entered the top 10 part of the
Top SSD Companies List
(compared to previous years).
2015 is my 20th year as an online web
publisher - and I've got plenty of evidence to show that search patterns for
technology can be correlated with later market adoption (and revenue) if you
know where to look, have the right data and understand what you're looking at.
top 50 SSD articles