| .. |
hostage to the fortunes of SSD"We can't afford NOT to be in
the SSD market...
But don't ask how we're going to make money out
of it yet.
(Or this year or next...)
Coz we're going to
do it anyway.
(This SSD stuff)"
by
Zsolt Kerekes,
editor - July 3, 2013
This is probably what any big server,
storage or chip company might tell you if they were being brutally candid and
weren't too worried about what was going to happen to their share price when
the statement hit the internet.
And it's what they might say if
they really understood what was going on in the SSD market - which I can tell
you from my own conversations and analysis - that very few people on the planet
- without the same kind of
crystal ball do
seem to understand fully - although as a reader of this site - you're already
much better placed than most.
But deceptively simple as my headline
appears - it only tells part of one strand of a complex market disruptive
story in which there are many plots, surprises and alternative endings - and
which - like those classic Russian novels - contains many characters who seemed
important at one stage - but whose name and role you can't quite remember when
you've finished reading volume 3. (Not everyone who starts - ever gets to
finish volume 1 - but somehow it became a classic anyway.)
here's
how this SSD article started
I'm often asked questions like
"when
is so and so SSD company going to be profitable?"
- or -
"why
are so many companies piling into the SSD market - when even the leading
enterprise companies haven't demonstrated sustainable business models yet?"
I
was dealing with a variation on that theme in one of several reader questions
yesterday. (When I reply to emails - as many of my victims can tell you - the
replies can be as long as some of my articles - and contain even more typos.)
But in trying to collect together 4 or 5 different themes in an email yestreday
- it struck me that I could summarize a lot of what I'd just said in a single
idea. Here its is.
Here's the real problem for the big companies.
They can lose tens of billions of dollars of revenue by not
participating in the SSD market.
This market is at the stage of "we
can't afford not to participate" rather than "we see clearly how to
make money".
And at the same time - one of the pressure points is... this is a
market where if you're clever and understand technology and how computers and
storage work you can still build a multi biliion dollar revenue business from a
small start.
That's where my email ended - and this was just going to be a quick
note on this web page before the holiday.
But if that's got you
wondering about the eventual ceiling of the SSD market (it's not a glass
ceiling - more like a cloud
ceiling) and who will get affected if they don't participate in the SSD market
experiment I suggest you take a look at some of the articles I've already
written listed in the
SSD analysts page.
Some
of these articles are very old. But they can still surprise those who haven't
seen these messages before.
For example it's 10 years since I wrote
that once the enterprise SSD acceleration market took hold in the user world -
the big server companies would all need to sell SSDs - because they were going
to sell 3x less servers - and no one was going to be impressed by the speed of a
server which didn't have SSD acceleration.
Unlike old jokes - old SSD
market models aren't always the best - even if some of them have surprisingly
well withstood the test of time. That's why there are so many on this site (SSD
market model related articles - not jokes - sorry.)
BTW - on a
historical
note - the company which did more than any other to change that old way of doing
server business was Fusion-io.
Now whether a server company hates or loves FIO - they have all become hostages
to the idea that they absolutely have to offer some kind of supported
PCIe SSD style
motherboad acceleration. They no longer have a choice in the matter.
Why?
Simple. If server makers stop offering SSDs - they'll sell even less servers
- because users don't see why they should pay to get a slow server at a fast
server price. But on the other hand if the server oem's favorite SSD supplier
gets
acquired by a
server competitor - they're vulnerable from that direction - because their sales
juice could get cut off. And they also need
more than one type of
PCIe SSD (the current flavor of server-side acceleration juice) to
satisfy all those different
use cases. So whether
they like it or not - the server companies have been sucked into an SSD juice
torrent where the only way they can stop drowning is to swim faster.
When
it comes to enterprise storage. We're still a few years away from the pivotal
point where 50% of enterprise storage on the
SAN will be solid state
rather than HDD based -
but some of the products which will prove the new
ROI metrics to
users are already shipping - and the comparisons will get better next year -
with better integrated SSD systems and software.
When it comes to the
semiconductor companies - the original driving force for
memory companies was
they knew the time would come when due to the complex characteristics of flash
memory in successive shrink generations - the only way to make a sellable
product would to add a controller.
Owning the SSD
brand (because users buy SSDs - not
flash memory chips) is
better business than selling commodity memory to SSD makers.
The
current source of pressure for flash makers is that they all believe that a
new type of memory design will be required in 2 -3 years to provide economic
manufacturing at smaller cell sizes.
But they all want to delay the
investment in new fabs (wafer fabrication plants) as long as possible until the
new ways of doing flash memory (or any
viable alternatives
to flash) are proven one way or the other. That's why for the next few years -
as they keep reminding everyone at investor conferences -
SanDisk,
Micron etc are planning
their product mixes around a predictable amount of semiconductor wafers based
on what they can do now with what they've got.
But that begs some
questions.
- How many SSDs can you make from a given known quantity of flash chips?
- Which SSDs will sell for more?
- How many of each SSD type can be sold?
You may think that sounds
easy - but even the first question has many different answers - depending on
your controller IP.
(That's one of the reasons SanDisk
bought
SMART. For the
details - see the SSD news
page.)
What about chipmakers like
Intel? - Well Intel
doesn't make flash any more - so it hasn't got exactly the same pressure as a
flash company. But the SSD pressure on Intel is acute for different reasons.
Remember what I said about users needing less server CPUs to get the
same jobs done?
In the near term that's going to shrink demand for
server CPUs. But more important than that - Intel - since about 1977 has seen
itself as being at the heart of microprocessor based system design.
And
for 3 decades most desktop computers and servers have been defined in relation
to how they compare with Intel's processor architecture and related operating
systems.
In the SSDcentric world - however - most of the dollars
which a user will spend on silicon chips will be in SSDs - and managing the SSD
assets will become the big focus of new
software.
In
that context the processor is irrelevant. All that software which
Microsoft,
IBM,
Sun/Oracle,
EMC etc have written around
the concept of fetching and writing packets of data to hard drives and
backing it up will
become irrelevant too.
Intel risks
being no longer be at the heart of the server design - in the same way as it
missed out on the mobile phone market. The stuff which Intel has been doing in
recent years with SSDs mostly confirms to my mind that the company still
doesn't understand the strategic importance of SSDs. In the meantime if you want
to know what the next Intel SSD looks like - just ask
LSI (and others).
I
could go on - but I won't - because this was just meant to be a brief note to
say - thanks for reading and thanks for sending interesting questions which
don't seem to have reliable answers. | | |
| . |
| hostage to the fortunes of
SSD - footnotes |
Editor:- September 12, 2013 -
When I published the above article on July 3, 2013 - it had been inspired -
like all my home page blogs in recent years - by many conversations I had
been having with SSD industry leaders and their investors and people in
organizations well placed to acquire any SSD companies they wanted. (If they
could satisfy themselves it would make sense.)
Underlying all these
confidential conversations is a desire to get a better understanding of what
has happened in the SSD market, what is happening now and where the SSD market
is going. There are many different views about how to interpret the same things
- as I've described in
an
earlier article.
In the months since writing "hostage to the
fortunes of SSD" there have been
acquisitions in the
SSD market by some of the companies discussed in it. Therefore I've kept
this blog alive longer than usual - because it helps you to understand the
general kind of thinking that's driving these activities.
In my next
blog I'll be talking about an upcoming market inflection effect I call the
"SSD software
event horizon" (which will dampen enterprise SSD revenue growth for a
while in some segments - but is an adjustment which will enable growth towards a
significantly larger SSD market in the very same affected segments soon
after) and also another 4 key technology factors - which you have to take
into account if you try to calculate how many SSDs will be needed to displace
all hard drives in enterprise servers, SANs and the cloud. (And put your own
date on when it will happen.)
The new article will also help improve
the resilience of your own SSD market projections or expectations and it will
also (I hope) give you the confidence to apply your own discount or scaling
factors to adjust any other SSD market reports or predictions which you see
elsewhere and to be aware of the flaws in them. | | |
| . |
|
|
| .. |
|
|
| . |
|
| . | |
.... |
| .... |
| Unlike before - today new
companies entering the market can rely on a sophisticated SSD ecosystem - in
which key elements of their solutions are already being supplied by other
companies. |
| what
changed in SSD year 2013? | | |
.. |
 |
..
.. |
 |
| .. |
..
..
..
.. |
|
|
..
..
..
..
..
..
..
.. |
 Was
there ever a Golden Heroic Age of the Enterprise SSD market?
and could
the plot of the enterprise SSD story have been rewritten with different
characters to make it simpler?
plot spoiler and
series review (2015 perspective)
|
..
..
..
.. |
|
|
..
.. |
 |
| .. |
|
| .. |
|
| .. |
|
.. |
| Many SSD marketers I talk
to nowadays regard the price per gigabyte of their latest product almost as a
badge of honor. But it wasn't always like that. SSD pricing used to be a taboo
subject. |
| SSD Pricing - articles
and trends | | |
... |
|
... |
|
... |
|
.. |
|
.. |
|
.. |
|
.... |
|
.... |
|
.... |
|
.... |
| .... |
| |
