|
by
Zsolt Kerekes
editor - July 2008 |
What effect will a
recession (if there is one) have on the
bubbling
solid state disk market?
This
is a serious subject - and before giving any answers I'm going to requalify
the question. |
What kind
of recession?
|
How bad?
If it's as bad as the recession in the 1930s - that's very bad for a
broad spectrum of industries (although the military industries seemed to do
quite well).
On the other hand if it's no worse than the 2000-2003 IT
recession which came after the dotcom bust, Enron collapse and 9/11 - that had
storage winners as well as
losers.
I'm going to assume for the purposes of this discussion
that if there is a recession in 2008/9 it will be more like the recent example
above - in which overall IT spending declined - 10, 20 or 30 per cent -
depending what you compared it to - but did not stop altogether.
Surprisingly
we haven't seen that kind of downturn yet in the rotating disk storage market
yet. An IDC report published in September 2008 - said that the external disk
storage market
grew 17%
year-over-year in the 2nd quarter of 2008.
But that doesn't mean
the storage market is imune.
With banks failing or merging on a daily
basis - there will soon be a flood of as good as new enterprise storage systems
flooding the broker / recycling market - just as when the dotcom bubble bust.
The impact may not hit new storage sales immediately - but users don't need to
buy new storage fresh from the factory if they can pick up the same products at
maybe a quarter of the price in an auction. |
| Which part
of the SSD market? |
Is there more than 1 part?
If you're new to SSDs You may think of "SSDs" as a single
segment. But there are a lot more. More than 2. More than 3. In the 5 years
since I started publishing various editions of my
SSD market
penetration model - which identified 4 main segments things have
gotten a lot more
complicated.
Let's just say that we're now well into double
digits - and every weeek this year has suggested a new way of looking at an
SSD market which in the long term will be bigger (in revenue) than the HDD
market is today. (Although the HDD market will
keep growing too.)
With
that kind of perspective - lumping the fate of all SSDs together makes about as
much sense as thinking about a single "storage market". Instead
(like investments) some parts of the market will go up while others go down.
|
|
 | |
A recession can kill, slow
down or accelerate the successful market adoption of new technologies. Which of
these happens depends on how technically delevoped the new technology is when
the recession starts to take effect.
My explanation of why some
technologies actually do better and grow faster in a recession is that
customers have less money to spend and have more time to research and think
about new upgrades and purchases - before they commit to a particular solution.
When enterprise customers have growing IT budgets (as in 2005 to 2007) the
easiest thing to do is buy more stuff from their existing suppliers and ignore
the distraction of qualifying alternative solutions. But when IT budgets are
crunched (as in a possible 2008/9 recession) user enterprises face the
dilemma of having to increase application performance (to stay competitive in
their own markets) while having less available funds to buy technology assets.
Buyers in a recession spend more time shopping around and researching best
value for money options. That's an opportunity for new technologies to do
well.
One example of switching away from safe (but pricey) solutions
in the last IT recession was the effective replacement of the SPARC Solaris
server market by Linux running on commodity Intel architecture servers.
Although
many
other factors were
involved too - the key effect of the recession was to speed up the move to
the new alternative. Linux had a much tougher time getting past the datacenter
door keepers during the preceding dotcom boom when the future success of new web
startups seemed to be measured by how nay Sun servers they had ordered.
Going
back to the subject of this article - here are my predictions for how the SSD
market will fare if there is a recession in 2008/9 - and the thinking behind the
conclusions.
- hybrid
flash disk market - will be a loser. This would have happened even
without a recession - because these product confer little or no benefit to most
PCs - while increasing the cost.
- flash
SSDs aimed at the notebook market. This segment will have winners and
losers.
Losers. - Products which have mediocre performance
(same or not much better than a notebook hard drive) will fail to be successful
- despite the fact that notebook SSD prices will drop as many of the 25 or so
manufacturers making such products try to gain market share or recoup
investments before exiting the market.
Winners. - Products
which are significantly faster than hard drives will be successful - because
there is always a market for faster products - even in a recession. Although SSD
oems who make such products wull see significant growth in units shipped - it is
unlikely that the high (double digit) penetration rates predicted by
IDC,
Web-Feet Research
etc will be achieved in the next few years - because the recession will put a
dampener on new notebook sales.
- flash SSDs aimed at the server acceleration market.
Flash
SSDs which offer significant application speedups are appearing in
2.5",
3.5",
PCIe and
rackmount form
factors. In many applications they will be more cost effective than a
traditional solution (more servers). And in some applications they will offer
performance which is years in advance of what current processors can perform.
But, an important limitation of many of these flash SSD enterprise products is
that most of them are optimized for a DAS rather than a SAN architecture. That
means their take-up is likely to be more in smaller to medium sized companies
or departmental rather than enterprise-wide applications in bigger
organizations.
Another market which has emerged in recent years is the
use of SSDs in appliances such as network switches. That too will accelerate in
a recession - because it's cheaper and more convenient for oems of such products
to offer a faster model as an upgrade (by installing an SSD to replace a hard
drive) rather than design or sell a completely new appliance.
- RAM SSDs aimed at the server acceleration market
With
RAM SSDs costing about
25x as much as flash
SSDs (for the same capacity) and the
performance gap
closing - you may think RAM SSDs are an endangered species - and that after
a recession this category will no longer exist. I confess that I too once
thought this might be the case. But recent detailed discussions with many
readers on this subject accompanied by much thought and analysis about the
various directions the SSD market could take - have led me to the opposite
conclusion. I'm now convinced that the RAM SSD market will be very much alive
and well after a recession and will actually benefit if there is one.
Most
current
RAM SSDs are designed
to suit a SAN architecture.
As the pressures to deliver increased application performance increase - but IT
budget cuts preclude the traditional (buy another 100 or 1,000 servers)
solution - more users will have the time to consider SSDs as a "distress"
upgrade. Deep pocket organizations with big databases - such as banks and
government have been doing this for years as recounted in many
case studies. But
these lessons have largely ignored by most user organizations - because when
the economy is good - it's much easier to ignore solutions that require more
thought - even if they cost less. RAMD SSD costs have declined considerably
since the last recession - which was probably the last time that many buyers in
big enterprises looked at this type of solution. (And then happily forgot about
them again afterwards.)
Another factor which will accelerate the need
for more "ultimately fast" RAM SSDs - particularly towards the end of
the recession and for some time after - will be the growing population of
installed DAS flash SSDs. A flash SSD accelerated server just looks
architecturally like a faster server.
When there are islands of fast
flash SSD accelerated servers - the only way to scale these applications up to
an enterprise level will be to bind the fastest common files into a RAM SSD
solution. That's what I call my "SSDs everywhere" model. More DAS
flash SSDs create more bottlenecks which can only be handled by more SAN RAM
SSDs. It's a kind of virtuous
performance uplifting circle in which neither technology replaces the other
- but the growth of one increases the demand for the other. Overall
a recession will be good for vendors of faster server oriented SSDs (both
flash and RAM based) - but a recession will be bad for vendors of notebook SSDs.
The
notebook
market was going to become a bloodbath anyway - because there are too many oems
chasing similar design slots with products which deliver little or no advantage
to the average user. The recession will speed up the fast-forward (or erasure)
of some of these market segments. |
|
| ......................................................................................... | |
... |
|
|
| ... |
the SSD Buyers Guide
After SSDs... What Next?
the Top 10 SSD Companies
RAM Cache Ratios in
flash SSDs
Storage
Market Outlook 2010 to 2015
3 Easy Ways to Enter
the SSD Market
the Problem with
Write IOPS - in flash SSDs
SSD Myths and
Legends - "write endurance"
RAM SSDs versus Flash
SSDs - which is Best?
Selling
revolutionary SSDs to conservative technology laggards |
| ... |
 |
| ... |
...Later:- April 24,
2009
...As we now know, there really was a recesion. But rereading
this article I wouldn't change a word of the analysis - which still holds true
today.
My conclusion (cut and paste from the bottom of this article)
hasn't changed either. Here it is.
"Overall a recession will be
good for vendors of faster server oriented SSDs (both flash and RAM based) - but
a recession will be bad for vendors of notebook SSDs."
For 2009
SSD market predictions - see
Looking Ahead to
the 2009 SSD Market | |
| . |
|
|
| . |
 |
| . |
| that RAM
SSD "Eureka Moment" |
Commenting for this
article Tomas Havrda, President
Solid Access
Technologies (one of the word's fastest growing storage companies) said
this...
"The recession is a good technology disinfector which
tests the products as well as business models.
"
Flash SSD
publicity helped make customers aware there are other storage subsystems besides
spindle-based devices.
They then discover that there are
RAM SSDs.
"
Customers are saying 'We can't keep doing what we are doing now'. It is not easy
to get into the door, but when customers actually try the ultra fast SSD there
is a "eureka moment" where they don't return our demo and
instead buy it outright." | |
| . |
| Z's Laws - Predicting
Future Flash SSD Performance |
A few months ago a
reader asked a very good question.
"Is there an industry
roadmap for future flash
SSD performance?"
That prompted other questions like...
- How fast are flash SSDs going to be in 2009?, 2010? or 2012?
- What are the technology factors which relate to flash SSD throughput and
IOPS?
- How close will flash SSDs get to
RAM SSD performance?
There wasn't a simple answer I could give at the time. Clues lay
scattered all across this web site
and in my many one on one discussions with readers about the market... But I
agreed there should be a single place on the web where these answers could be
found.
Forget
Moore's
Law. That gives you the wrong answer, and this article explains why. ...read the article | |
| . |
| Where's the
Money for SSDs Going to Come From? |
There's no such thing
as "new money" in an IT spending slowdown.
Here's an
editorial first published on our home page back in January 2008 - which
basically says - more spending on SSDs will mean less spending on something
else.
2008 SSD Budgets - Robbing Peter to Pay Paul?
I've
got a question for you. How big is your
solid state storage budget
in 2008?
I hazard to guess that most of you haven't put anything
notionally marked as "SSD spending" in 2008's budget - even if you
already have a good idea about what you're going to spend on traditional storage
products and services.
I also predict that when the crunch comes - and
you find yourself spending surprisingly large amounts of money on SSDs for the
first time - these costs will be initially allocated to other cost centers -
such as servers or PCs - rather than storage.
It was always thus.
In 1983 for example - over 90% of corporates didn't have a budget for buying
IBM PCs. These disruptive tools intitially crept in under the IT department
radar - as users found they could do useful jobs like word processing and
business analysis quicker, cheaper and more conveniently than using the clunky
alternatives then on offer by their IT departments. Similarly
RAID systems did not
appear in most 1990 corporate IT budgets - but are now everywhere.
In
2007 the SSD
industry surprised many by introducing many exciting new technologies and
products.
I predict that in 2008 - innovative users will surprise the
SSD market by discovering for themselves a new generation of killer
applications- enabled by SSDs - which would have been technically impossible -
or even nuts to try and achieve using conventional
hard disk based
technologies. Those SSD sparks will feed back to fan the flames of the market.
Examples might include what I call - enterprise spreadsheet analysis -
in which business managers are enabled to model "what if?" scenarios
on duplicate sets of their entire customer database - to find gaps in their
marketing or test ideas for new products.
Other applications enabled
by SSD accleration might include AI enabled real-time upselling offers on
ecommerce web sites. These are already feasible for innovative small to
medium size companies but are not scalable with today's magnetic disk arrays.
Large enterprises can't be nimble with their data because their servers would
grind to a halt if you tried these types of tricks.
I'm not going to
give you a long list of predicted disruptive SSD technology enabled applications
- because they would be wrong - and you're going to see them coming thick and
fast from real users on these pages soon enough.
2008 will be the year
that users - rewrite the rules on how they mix and match new storage
technologies in ways that the original manufacturers of those products - never
dreamed about. | |
| . | |
