market sizing the architectural impact of "SSD
10 years after publishing this
article - market
that 2003 did indeed mark the start of the
Modern Era of SSDs
trust SSD market data?
what's the big
picture message re SSDs?
5 User Value
Propositions for buying SSDs
are we ready for
infinitely faster RAM? (and what would it be worth)
|could enterprise SSDs
become a $10 Billion Market?|
I was recently asked by one
manufacturer for the original source for STORAGEsearch.com's frequently quoted
editorial view (in 2003) that the Solid State Disk market could reach $10
Billion per annum. I am the source, and this article reveals my thinking and
the evidence supporting that view.
Historical Data Which
Supports the SSD Market Size Model
There has been significant growth in reader interest in
solid state disks. In Q1 2001 - SSDs were the 18th most popular subject. In Q1
2002 - SSDs were #4. In Q1 2003 - SSDs were #2. In Q2 2003 - SSDs were #1.
That means more readers than viewed our
NAS page for example.
Clearly something has been happening in the market to accelerate reader
interest. That was the first clue.
Our readership tracks the
interests of hundreds of thousands of people who looks for storage focused
content on the web. Our 11 year history as an IT directory publisher, strongly
indicates that changing reader interest in subjects is a strong predictor of
future market growth. It can take several quarters for that revenue to occur,
and upto a year before it is reported by market research organizations which
track revenue. But during 8 years of tracking these trends online in the
storage markets, we know that the
effects are real.
As a sanity check - consider that
Serial ATA was the #1
product category on STORAGEsearch in the 2nd half of 2002 - clearly predicting
future interest in a market, which at that stage didn't have many real shippable
products. It's the changes in these rankings which are important. Some
categories stay in the top 5 year after year. But when products get into the top
5 from relative obscurity, that signals volume changes are about to happen.
other Market Researchers?
There isn't any data being
collected about SSD market size.
Companies like Gartner
track removable USB flash drives. But these flash storage modules aren't
the same as true
SSDs - and they exclude entirely
DRAM based SSDs.
Many SSD companies have disclosed to us that their
own revenue growth has been significantly higher than 100% year on year.
Looking Ahead - Market Trends - Which Supports the SSD Market
introducing a new concept in computer architecture -
SSD CPU Equivalence
Our analysis at STORAGEsearch.com is that
the correct market size for SSDs (used as server accelerators) is on the order
of $10 billion The factors which will drive this are as follows
1 - users in the early adopter category have discovered that within
their IT budgets they can increase application performance at lower cost by
switching spend away from server CPUs over to SSD accelerators.
example customers looking to upgrade heavily utilised 60 CPU Sun systems can
choose between doubling CPUs by buying new servers or spending about 25% of that
server cost on SSDs. Other users, lower down the scale, have been reported as
taking servers out of use to reduce software costs, and using SSDs to make up
the performance difference. Therefore the SSD accelerator market should be
viewed as a replacement for part of the server CPU market. Not as a
percentage of storage spend.
2 - Sun and others in the 64 bit server market have already seen
decreasing returns in server performance with higher GHz CPUs. This trend will
get worse when new generation Intel and Sun processors, which use more CPUs per
chip come into mid range products. Future growth in server performance will not
be possible unless server OEMs start to architect their systems to include SSDs
as a standard option.
3 - Historically the SSD market has been limited by poor software
tools. Users were unable to get promised speedups without extensive technical
knowledge. Recent software tools from Imperial, BiTMICRO and others hold the
promise that users or VARs can tune their own systems. Therefore
sales ramps will no
longer be limited by application engineer headcounts in the SSD companies.
4 - upto Q2 2003, the major server oems, IBM, HP, Dell and Sun have
not been involved in any significant way in the SSD market. But when markets
reach a critical size, the big companies will step in to protect declining
server revenues. This will most likely start with the acquisition of SSD
companies and OEM badge agreements. The effect will be that SSD technology
will be marketed by companies which are more familiar to users than the SSD
pioneers. These factors will accelerate the market growth and channels to market
for SSD technology.
Extrapolating from the current market
size, and with the potential to replace other products (such as server CPUs and
application software licensing), it is reasonable to assume that the trends now
seen in the SSD industry are part of a disruptive change in the computer market
which could propel Solid State Disks into a $10 Billion Market by the end of
Phase to a Disruptive Market in Solid State Disks |
|It's users who create disruptive markets,
not product marketers. If enough users think that a new technology will fly - it
will take off. Conversely there are many new products which marketers think are
a good idea, but which fail in the market.|
The high interest in SSDs
on STORAGEsearch shows that a significant number of users (over 100,000) have
been looking at the idea with interest.
The transition phase to a mass
market will happen something like this...
Applications engineers in
large server companies like Sun, HP and IBM will work with 3rd party SSD vendors
in niche markets. Product marketers within the server companies will learn about
the new technology and make it a standard "special" supported
feature. Competitive pressure to improve the benchmark ratings of 64 bit multi
processor chip based servers (like the SPARC IV) which have hit glass ceilings
on Gigahertz bus rates will lead to an oem including SSDs in their basic server
offering. This will double the published benchmark performance.
3 months - every major server manufacturer will do the same. That's when the
acquisitions and serious partnering with SSD companies will begin and we will
switch to a mass market. Initially server companies will badge engineer SSDs.
Eventually technology licensing and acquisition will mean that most will make
their own. There will still be a non captive market for SSD companies as
upgrades and working with smaller sever oems just as there is today with RAID
The process could be accelerated by flash semiconductor
companies who will see SSD revenue as a way to make up for not having the
processor crown jewels (that is not being Intel.)
This is similar to
what happened with RAID systems which started out as niche add-ons in the late
1980s and took nearly a decade to become commonplace. The 2000-2003 US IT
recession meant that users didn't go out and buy many systems that they didn't
absolutely need - but the absence of a large visible market shouldn't be
confused with indifference. many technologies including Disk to disk backup,
iSCSI and SSDs were being evaluated or considered during this period and will
change the landscape of the storage market in teh recovery. See
article:- Storage Winners
and Losers from the 2000-2003 IT Recession.
One factor which SSD
product marketers have to remember is that users behave in a way which reduces
their risk. Even if they like your technology they may not like the idea
of buying it from your company. But they will feel more comfortable about buying
your product if it's supplied by a company they trust - like a reseller they
know - which is selling the product supported by the company which makes the
Losers from a $10 Billion SSD Market?|
predicting which companies will be the biggest winners and losers, when new
technology disrupts high tech markets.|
Currently most SSD companies are
benefitting from these market trends, but many lack the marketing skills
required to take advantage of a much bigger market:- development of
software tools, VAR
etc. Some pure technology companies can only
hope they get
acquired, by the big server companies.
Intel will lose server CPU
sales, compared to today's market patterns. But servers are only a small
percentage of Intel's overall processor sales. On the plus side, Intel could
stand to gain much more revenue from increased flash memory sales.
Oracle and other enterprise
software companies will lose the most. Because users will deploy a lower number
of server processors than they do today.
About half the number of
processors could perform the same tasks in a typical organization, with the
benefit of SSD and other network accelerator technology.