Updating my SSD market adoption and
explaining - why users will buy SSDs
by
Zsolt Kerekes
editor - StorageSearch.com -
November 2005
This is a long article - which provides a framework
which will help you understand SSD market adoption upto 2020. Here are the 6
SSD market cases discussed.
- hostile environments
- server acceleration (SSD-CPU equivalence)
- road warrior and featherweight notebooks
- high reliability embedded
- low cost SSD bulk storage / enterprise HDD array replacement (added 2010)
- tiered DRAM replacement in servers (added 2015)
See also this
(later) article:-
are we ready
for infinitely faster RAM?.
|
User Value Propositions for
buying SSDs
the original text which was published here on
StorageSearch.com - November 2005
I was the first storage
market analyst to publish a report saying that the enterprise SSD market had
the potential to reach $10 billion annual revenue. I described the
economic and technical factors in an article in
2003 (when the
entire SSD market revenue was under $100 million).
So why are most
IT editors and analysts today (in November 2005) still so wrong about
the potential of SSDs? - and prefacing every discussion of SSDs with a
superficial analysis which compares the cost per byte of storage between
flash and
hard disk drives?
There's
a simple explanation.
They really don't understand the solid
state disk market.
Magnetic versus flash media cost comparisons
provide the wrong answer to the wrong question. And are far removed from why
the SSD market is racing to become a multi billion dollar market seemingly in
blithe ignorance of the cost per byte proposition. This article tells you what's
important to users and the main applications in which SSDs are already being
used and new applications where they will be used in the future.
|
STORAGEsearch.com is
recognised by most SSD
manufacturers as a leading authority on recent developments and changes in
the emerging SSD market. This article updates our earlier SSD market penetration
model, first published in
2003. It also predicts and describes the emergence of some new
applications and markets.
I first started using SSDs in computer
systems over 25 years ago, and have tracked the market as a publisher
for over a
decade. The STORAGEsearch model and insights presented here are based on our
own original market research, discussions with SSD oems, analysis of reader
trends, discussions with readers and reading (or publishing) much of the
original literature related to this market in the last 5 years.
Before
presenting our market model - let's go back to the premise in this article's
title above - "Why are Most Analysts Wrong About Solid State Disks?"
One
simple answer is that analysts and market researchers have nearly always
failed to anticipate, recognise or predict the emergence of new disruptive
technology markets. This phenomenom is well documented in marketing text books.
The
SSD market is no different in that respect to any other disruptive market.
Analysts tend to be good at predicting incremental changes in established
markets where there already is plenty of data about revenue, users and product
shipments. It's expensive to collect that data, and someone has to pay for it.
In the high-tech market when product niches are below about $500 million (using
2005 values) - it is not economic for traditional big name market researchers
to collect any data. In the absence of sponsors and market research which asks
users why they are buying new technologies - market research companies fall
back on using traditional analysis techniques - usually applied to the wrong
parameters.
In the SSD market - the wrong parameter to look at is - the
price per byte of solid state disk storage (flash or RAM
price) and compare that to the price of rotating magnetic media in hard disk
drives. After quoting publicly available price points - the
sages wisely pronounce
that the projected graphs may cross and reach parity at some distant time in the
remote future but the time has not yet come - and they can't see what all the
fuss is about.
Before progressing to our own market penetration model -
I thought it would be instructive and amusing to see how that kind of
traditional analysis - which looks at the wrong data from the wrong point of
view - would have predicted the outcome of some earlier disruptive
technologies:-
- the car:- was a disruptive technology in the 20th century which
mostly replaced walking, the horse and the train as a method of transport. Yet
if you look at the cost to buy, or the cost to own, or the cost per mile - then
the roads today should be filled with motor bikes and not cars. Cars cost more.
Are we all nuts?
The customer value proposition is that cars offer
greater flexibility and comfort in personal transport. Even though cars are
mostly used by a single driver, they offer the flexibility of being able to
carry more passengers. And even though motor bikes offer lower cost when
carrying 2 adults (and 2 bikes carrying 4 passengers costs less than a car
too) most users still prefer to buy cars for everyday transport. (Bikes and
their leathers, in the US and western Europe have become a fashion / lifestyle
statement - but that's outside the scope of this discussion.)
- the original IBM PC:- the killer app for the early PC market was
word processing. But in the early 1980s a PC with printer cost about 20 times as
much as a portable typewriter and about 4 times as much as a heavy duty electric
typewriter. PCs were also more complicated to operate. Looked at on a pure cost
to buy basis - the PC market should never have taken off in the document market.
However, as we all know - typewriters became extinct. Who could have predicted
that?
The customer value proposition was that users could standardise
documents and the flexibility of PCs was judged superior to that of typewriters.
Although in a rear guard action - office based typewriters did start to sprout
floppy drives and editing screens - the PC is the tool which won.
- the Apple iPod:- why would any sane person in 2004/5 buy an iPod to
play music - when a portable CD player or cassette tape player costs less than
1/4 of the price - and doesn't need a PC to set it up?
As we know,
tens of millions of users have bought iPods and the market for similar
technology portable music players will soon pass the hundreds of millions mark.
So the cost per player is not the important issue in the minds of the user.
The
customer value proposition of the iPod is that it delivers a more flexible and
rewarding entertainment experience compared to listening to the same small
number of CDs (or tapes) over and over again until you are bored, or the
cumbersome alternative of carting around a crate of music media. Let's
move on to the SSD market and see why cost per byte analysis has failed to
predict or explain the emergence of a multibillion dollar SSD market. |
| STORAGEsearch.com SSD Market Penetration Model (Nov 2005) |
| Type |
Application |
Narrative |
| 1 |
Hostile
Environments
|
This was the earliest use of
SSDs.
In the late 1970s manufacturers of industrial control systems
used solid state disks to hold programs - because the hard disks available at
that time were expensive and unable to operate reliably in a factory environment
(vibration, temperature and power fluctations being the main stress points).
For similar reasons SSDs were used in military embedded systems. Competing
with SSDs specialist military manufacturers designed cannisters which reduced
the amount of vibration transmitted to hard disks, enabling their deployment in
some vehicle and mobile applications, and the durability of HDDs improved during
the 1990s. But there are many environments where HDD media itself cannot survive
- such as extremes of high and low temperature. By the end of the 1990s flash
solid state disks had started to replace HDDs in most miltary applications due
to superior operating temperature, lower weight, lower power and faster
performance.
Today one of the iconic applications for mobile consumer
storage is portable music players such as Apple's iPod - in which HDD or flash
SSDs are used depending on the price point. But much higher volume markets where
SSDs will dominate, and where mobility and environmental factors are important
include:- cell phones, in car entertainment and navigation systems and cameras.
The
customer value proposition in Hostile Environments is that SSDs operate in
environments in which hard disks are unsuitable, and where SSDs have superior
weight, power, reliability or other key attributes.
An excellent
discussion of how the "floor price model" affects the interplay of HDD
and flash in consumer applications is presented in this
article:- Flash Memory
vs. Hard Disk Drives - Which Will Win? written by Jim Handy
Semico Research |
| 2 |
Server
Acceleration
|
As long ago as the mid 1980s,
engineers (like me) using SSDs in
military systems for
type 1 applications (above) noticed that they also got speedup benefits from
using SSDs compared to using HDDs.
Some
RAM based SSD products
were launched in the 1980s specifically to offer speedup (instead of
environmental survivability). But as hard disk interfaces got faster, and cache
became more common - most early SSD products aimed at the server acceleration
market had died out by the early 1990s.
In the period 2000 to 2005 -
SSDs started to become cost effective and viable in high end commercial server
applications. In the next several years the arguments for using SSDs for server
speedup will become more compelling and the market will change from SSDs being a
rare technical deployment in 2005 to being a commonplace item by 2008.
One
way of thinking about this concept in computer architecture is - SSD CPU
Equivalence.
For a wide range of applications if you take a black
box approach and analyze the overall application performance of a computer
system - you would not know whether that system had more CPUs with hard disks or
less CPUs with more SSDs.
In most multi-user database driven applications today the factor which
limits performance is not server MIPs but random disk access speed. In the past
20 years disk throughput (Megabytes per second) has increased by a factor of
x100 (from 1MB/s in 1986 using 5.25" SCSI disks to 100MB/s sustainable in
2005 using 3.5" SAS
disks) but the random access time has improved by a much smaller rate of only
x5 (3,000 RPM disks in 1985 versus 15,000 RPM disks in 2005.)
In the
same 20 year period - processor clock speeds have risen from 33MHz to over 3GHz.
The data bus widths have increased from 32 bits to 64 bits and new processor
chips include from 2 to 8 internal processors. That's a x400 to x1,600 increase
in data demand for a typical server.
Many users have discovered that
when they need faster performance they reach a bottleneck or ceiling which
cannot be improved by simply adding more processors or memory. But by adding
high speed SSDs to critical storage segments they can get application speedups
from x2 to x40. SSDs can deliver more than 1,000 times the random access speed
of hard disk based systems. If deployed correctly users can speed up their
applications and save money by neeeding less servers, less software licenses and
less systems to support.
The customer value proposition of
Server Acceleration is that SSDs double the speed of enterprise server
applications at a price which is much less than buying another 10 to 100
servers.
The user adoption of this technology has been slower than the
technology permits for the following reasons
- most users are not computer architects, and have no idea of what
performance they should expect from their systems
- server manufacturers have a disincentive to promote knowledge about SSD
accelerators - because this would result in them selling less servers - and
getting a reduced revenue stream
- most SSD vendors have been small or medium sized companies which didn't
have the marketing muscle to educate users about the technology
In the
present market - this type of sale is a technical sale - in which a user hits a
performance problem which can't be solved by their server supplier. In
desperation and in a sceptical frame of mind they evaluate SSD technology - and
are usually amazed by the results.
65% of SSD users surveyed
said "It greatly exceeded expectations - I advocate others to try"...
from the SSD Buyer
Market Preferences Report)
The transition from a niche to a
mainstream market will be helped by the following factors.
- Better user awareness - due to publications like this one, and better
marketing and education by SSD vendors.
- Server oems will start using SSDs as an engineering fix for big customers
who are disappointed by the performance of new generations of servers. The
mismatch betwen server IOPs and hard disk IOPs will become more apparent with 2
core and 8 core processor chips. Customers will want to know why their new
servers aren't any faster than their old ones. Eventually, faced with declining
server volumes, server oems will start to see selling SSDs as a revenue growth
opportunity instead of a revenue reducing one.
See also:-
RAM versus Flash
SSDs - which is Best? |
| 3 |
Road Warrior
Featherweight
Notebooks
(weighing under 1kg / 2lbs)
|
This market has not started yet
(in 2005) but will kick into play during 2006/7.
My wife is a marketer
who travels a lot by plane (in Europe) and train (in London). Running Powerpoint
is critical - but so is the ability to pack a notebook PC and a week's worth of
clothes and paperwork into a lightweight bag.
Notebook manufacturers
like Sony, Toshiba and Dell love this kind of customer. Her lightweight
notebooks typically cost 3 to 5 times as much as similar looking luggables
weighing 3 to 4 times as much. And she changes them every year - because they
never quite have enough performance.
In this part of the notebook
market customers pay a hefty price to get less weight and better battery
operation... And adding insult to injury - low weight notebooks also have
processor clock speeds which are typically 3 times slower than desktop PCs (or
luggable notebooks). The expensive notebooks also have slower hard disks
because that's another way manufacturers deliver longer battery life.
What
can flash SSDs do for this market?
A 30GB
flash solid state disk
can act as a speedup accelerator - complemeting a low power hard disk - so
that a 1GHz lightweight notebook processor delivers similar performance as a
3GHz desktop model. (As CPU clock rates rise - the benefit delivered by the SSD
actually increases.)
How much will high powered road warriors be
prepare to pay for a flash disk which makes their featherlight notebook PC run
as fast as a desktop or a luggable?
$2,000? $1,000?
You'll see
that the cost, compared to a hard drive is not the relevant factor.
The
customer value proposition in the Road Warrior Featherweight Notebook
market is that the SSD provides desktop application performance in a low
weight, long battery life form factor which is impossible to achieve using
microprocessor technology. (Where high speed - means high power, fans etc.)
As
flash disks are on a steeply declining cost curve - and new entrants to the SSD
market apply learning curve pricing this application segment for SSDs will grow
to billions of dollars in the next 2 years.
...Later:- as
predicted - notebook
SSDs did become a significant new market - although the road to adoption was
much slower than originally expected due to many badly designed and poorly
integrated products along the way. |
| 4 |
High Reliability DAS
and
Infrastructure Blades
|
Although the superior
reliability of SSDs has always been a factor in hostile environments (see 1
above) there are new SSD products which have been cost engineered to replace
disks in environments where hard disks actually operate without too much
trouble.
This is a conceptually new market - pioneered by
SiliconSystems
in 2004. But other flash SSD makers are also moving in too.
The
application is using high reliability flash SSDs to replace disks in servers to
reduce service calls and service timeouts. Some of the markets where this
approach can be used are in embedded systems in telecoms, cell phone hubs,
server farm blades, and physically distributed infrastructure machinery which
traditionally uses hard disks. This kind of embedded application often uses
much smaller capacity disks than PCs.
The customer value
proposition of the High Reliability DAS SSD is that the interval between
server failures will be extended by several years compared to HDD technology.
Since the cost of sending out an engineer to a remote cell phone hub
(say) to replace a hard disk is orders of magnitude more expensive than the
cost of the hardware - the SSD delivers much lower cost of ownership.
This
type of operating model suits utilities and other industries which have
geographically dispersed servers. But it can also apply to server farms in
datacenters too. Flash SSDs not only provide better reliability, but much lower
electrical power consumption and lower cooling costs too.
At first
glance - all flash SSDs offer better MTBF (mean time between failure) than hard
disks. But new design concepts and data management algorithms inside the SSDs
means that some vendors' flash disks offer operating lives which may be 2 to 3
times longer than others. We are still in the early days of this sophisticated
new market - which is far removed from the consumer market and typical corporate
datacenter. But as the capacity of these high reliability disks rises and the
cost benefits become better understood - we could see the concept moving into
more more traditional IT markets too. |
| 5 |
SSD bulk storage
lowest
TCO for multipetabyte backup, archive and seldomly accessed data (low DWPD)
in the cloud
even if hard drives are free!
|
This is a new add-on to
the SSD market adoption model - added 7 years after the 1st edition and 5 years
after the text above.
Editor:- March 16, 2010 -
StorageSearch.com
published a new
article which describes the roadmap for the barely nascent
SSD Backup Market
to replace the enterprise
hard disk backup market by
the close of this decade.
There will be many technology and
marketing challenges along the way. It will require entirely new types
of SSD products and new ways of thinking about what the purpose of
backup really is. You
may be thinking - This can't be serious! Is it April 1st already?" You
may change your mind when you've read it. ...SSDs - reaching for
the Petabyteyears as they saw product examples in the market |
| 6 |
low cost DRAM replacement in
server |
The concept of replacing a
large proportion of DRAM in
servers (and reducing server counts again) with flash or other low cost (but
high capacity memory - compared to DRAM) started to become apparent from the
user experience of server based acceleration in the
PCIe SSD market. The
articles below discuss the thinking and products related to this trend.
| |
Summary
SSDs
are seeping into many market applications long before the time that would be
predicted by analysts who cling to the outmoded model of cost per byte parity
with hard disks. This article has shown what I believe will be the biggest
markets for SSDs in the next 2 to 3 years. I will update the model from time to
time - and in the meantime keep tuned to the news and other developments as they
appear on our main SSD
page.
My thanks to all the people who have helped create this model by
their inputs, discussions and questions. I hope you too Dear Reader, find it
useful. As with all models which predict emerging markets - it is prone to being
wrong. But it's the best I have to offer at the present time. |
|
|
| . |
In September 2010 - I
published a new series of articles on the theme -
Branding Strategies
in the SSD Market - which includes case studies describing how various SSD
companies have used branding to stake out their turf in the expanding
SSD market bubble.
In
March 2012 - I published an introduction to
enterprise SSD silos - It describes 7 SSD types which will satisfy all
future enterprise needs.
In May 2014 - In my article
Decloaking
hidden segments in the enterprise for rackmount SSDs I noted that despite
all the investment and
marketing activity which had taken place in the enterprise SSD market - and
despite the growing attractions of this as a place for vendors to focus - there
were still many types of customer requirements which were being ignored or
not serviced by suitable products. | |
Retrospective view from
SSD market
history
This is the original SSD market adoption
model which was adopted by SSD product marketers worldwide and helped to ignite
the industry!
I know from the many contemporary discussions I had
with SSD product marketers and company founders that the ideas and models
discussed here - helped those companies see the potential of the SSD market in
a new way - which many of them went on to incorporate in their business plans. | |
| . |
 |
| . |
 |
| . |
| Solid State Disk Buyer
Market Survey Results |
Editor:- in Q4 2004 -
I designed and ran the world's first major market survey designed to learn
more about SSD buyers needs and preferences.
It was my idea to do the
survey, and to keep it strictly independent I I didn't discuss the design of
the questions with any SSD vondors and I didn't ask for any sponsorship.
This
article provides a summary of highlights from the survey results.
The
survey has identified technical gaps which require new product solutions and
service gaps which require changes in the marketing plans of SSD vendors who
need to change the way they do business.
SSD vendors must take note of
the signals flagged in this survey if they wish to transform this market
segment from a niche technical market into a mainstream multi billion dollar
pillar of the storage market. ...read the article | | |
| . |
| SSD Pricing -
where does all the money go? |
| SSDs are among the most
expensive computer hardware products you will ever buy and comprehending the
factors which determine SSD costs is often a confusing and irritating
process... |
 |
...which is not made any
easier when market prices for apparently identical capacity SSDs can vary more
than 100x to 1!
Why is that? ...read the article to
find out | | | |
| . |
 |
| . |
| this way to the Petabyte
SSD |
In 2016 there will be
just 7 types of SSD in
the datacenter.
One of them doesn't exist yet (in 2010) the bulk
storage SSD.
It will start to replace the last remaining strongholds
of
hard drives in the
datacenter due to its unique combination of characteristics, huge storage
density, low running costs and operational advantages.
Bulk
storage SSDs will displace the last remaining hard drives in the enterprise
server market by 2020 - even if the price of a new hard disk drops to
zero and enterprise HDDs are given away free! |
 |
... |
The new business and
architectural models of the datacenter - how we get from here to there - and
the technical and problems which will need to be solved - are just some
of the ideas explored in this
visionary article. | | | |
| . |
| the future of enterprise
data storage? |
Editor:-
the
future of data storage is the lofty sounding but aptly chosen title of a
new article published in the January 2011 edition of Broadcast Engineering
magazine.
It's written by Zsolt
Kerekes editor of StorageSearch.com
(that's me).
It's a completely new article which takes as its starting
point - storage market models and concepts from several futuristic
articles which have already appeared here on the mouse site - advances them
and integrates them into a single cohesive whole. |
 |
It will give you a clear
idea of how all the incremental changes you read about in
storage news pages will
add up to a different future - and the business reasons why. Sit back and
...read
the article | | | |
. |
 |
.
.
.
.
.
.
. |
|
|
.
.
.
.
.
.
. |
| |
