moving from the language of storage
speed to Nx apps speedups
Moving the language of SSD
apps acceleration closer to something which is meaningful to customers - some
enterprise SSD marketers will move away from quoting discredited
and closer towards promising actual Nx speedups for specific popular
...A few weeks after publishing this - an amusing blog
was published on the same theme - "One
million IOPS. Yawn! Is that all you've got?"
Speed and Strength
Metaphors in SSD brands
||plunging deeper into MLC flash SSD babble is enterprise's best
Continuing to satisfy the needs of technically
minded SSD users who have already digested everything there is to know in the
2010 version of
SSD jargon dictionary
- the SSD industry (including this editor) will talk in a lot more detail about
nuances within MLC flash technology, such as the differences in different types
of so-called "enterprise grade MLC flash" and why some MLC SSDs are
nearly as good as SLC (as long as you don't cook them) while other sorts of MLC
are only good enough for
(and sometimes not
even good enough for them).
...Later:- these market and technology
trends were confirmed during 2011 and are discussed in the article -
sugaring MLC for
||enterprise SSD market - growing faster than
analysts - (in the terabyte tallier category - that is to say - who publish
reports on how big the various segments of the market are with respect to
historic revenue) will reveal to everyone's "surprise" that the
enterprise segment of the SSD market grew much faster in 2010 than they had
predicted - whereas the opposite case was true for notebook SSDs.
back on 2011 - there was a flurry of new investor interest in the enterprise SSD
market which arose from Fusion-io
going public and SanDisk's
acquisition of Pliant.
These taken with the details revealed by various SSD makers about how many
petabytes of SSD
storage they had shipped - confirmed that the enterprise SSD market was
moving into uncharted revenue regions faster than previously anticipated.
my discussions with leading enterprise SSD companies - the underlying reason
appears to be when when a big user organization has tested and confirmed the
benefits it gets from its first few cautious installations - the customers
become more confident about what they can achieve with SSD acceleration in other
parts of their business.
||SSD accelerated disaster recovery|
StorageSearch.com will publish a news story - "Business Saved by
SSDs!" - about an online enterprise which recovered from a disaster
recovery situation in just a couple of hours (accelerated by SSDs) instead of
several days (using traditional hard disk technology and tape).
case study will shine a new light on
what SSDs can do for backups
- and will help to open up a new strand of the SSD market.
although we haven't published such a story yet - many SSD vendors have told me
that one of the benefits their big enterprise customers are seeing is faster
backup and restore.
blog by SANRAD said - "...You can nearly double the utilization of
your flash resources by sharing the stored data for the dual purposes of both
volume replication (protection) and volume acceleration (performance)."
||Auto tiering SSDs / SSD ASAPs|
In 2011 we'll see some
clarity emerging in the SSD
ASAP market (auto tuning SSD accelerators which work alongside legacy hard
disk arrays). The attractiveness of this concept is that while SSDs remain
too expensive to replace HDDs and there's a shortage of the expensive technical
talent which can tune
apps for SSD acceleration - these black boxes can in theory deliver 2x to 3x
speedups by learning data hot spots and adapting their caching accordingly.
2010 was a
phoney war in the ASAPs market - because they were
too new and
unproven for users to trust them. So vendors had to fight hard to get any
decent beta test sites with live customer data.
In 2011 - the year
of reality checks - vendors of SSD ASAPs will have to refocus their
marketing efforts on specific types of applications where their boxes work well
- and avoid wasting sales resources on applications where their algorithms don't
work at all - and even make things worse.
There will be attrition too
- because (based on some of the benchmarks I've seen and reported) some of
these ASAP solutions are uneconomic and don't deliver any speedup benefits at
all compared to not being installed. On the other hand - those which do work -
could start to make a lot of money if they are promoted effectively.
||notebook SSD market |
market had a
murky 2010 -
and in 2011 the range of options being offered to users will continue to
puzzling and bewildering.
At one end of the spectrum - companies like
Seagate will continue
flogging what I believe to be the dead horse of
hybrid drives -
insisting that just because this concept failed to delight users in previous
manifestations - this time the breeding process has resulted in a product which
will work and negate the need for users to transition to pure SSD storage.
new alternative to hybrids - may come from
SSD software vendors
who are suggesting that
auto tiering within a
PC using a vanilla SSD alongside a vanilla hard drive will provide a better
balance of speed, capacity and cost - than all-in-one integrated hybrids.
Meanwhile the SSD market will continue offering a well stocked cocktail
bar of colorful SSD juices with different form factors, capacities and
interface types and hope they can spot the mixture which appeals to most
palettes. Adding to the obvious product appeal of performance, price and battery
life - notebook makers will also start to integrate new "SSD only"
features into high end products.
military style security features in a notebook SSD in Q1 2010 - in which the
SSD can self erase if
it is stolen. But integrating new concepts seems to take much longer in the
PC market than in the more dynamic phone market - where there is more OS
||Data Integrity glass ceiling rises to
the primary incubator for flash SSDs was the
military market - in
the past 5 years or so most of the pace in architectural innovation in the flash
world has come from companies in civilian markets. But one surprise for me in
the military SSD market in 2010 was the combination of features in a new
2.5" SSD called the
Guardian from Microsemi.
As you'd expect in this sort of product it's got
fast purge, and is
stable during power fluctuations without the use of super capacitors or
batteries - but it also pushes the boundaries for data integrity - specified as
an uncorrectable bit error rate: better than 1 sector in 1030 bits
read. Although some vendors in the civilian enterprise SSD (such as
have already started talking about the
to end data integrity offered by their SSDs - Microsemi's Guardian pushes
data integrity specs to new limits. It shows that we can expect a new class of
ultra high data integrity SSDs to emerge in the market in 2011, 2012 which
start laying the foundations for
||RAM SSDs - growth accelerated by flash|
number of vendors in the RAM SSD market will grow in 2011 - and so too will
revenue in this segment.
This is the opposite to what many people in
the SSD industry themselves would have expected 3 to 5 years ago when some
analysts were predicting that flast flash SSDs would kill the RAM SSD market.
The reasons why the RAM SSD market did better than most people expected in 2010
- and will do even better in 2011, and 2012 are analyzed in exclusive articles
and reports on the RAM SSD
||PCIe SSDs - not a clone market |
The high market
PCIe SSDs has set new
performance and price expectations which cannot be met by interfaces
originally designed for HDDs
such as SAS and
one bus - and speed as the over-riding reason for using it - you might think
all PCIe SSDs would blur into a sameness of functionality - but there were
already signs in 2010 that wasn't going to happen.
The sub groups and
trends within the PCIe accelerator SSD market are described in the article -
don't all PCIe SSDs
look pretty much the same?
||rackmount SSDs - more sub-groups |
market will continue to see
in architecture and the definition of what makes a good rackmount SSD.
This will make it harder for users to recognize what products might work for
them - due to the morass of technical data in product descriptions which they
have to wade through to filter relevant products.
One approach we'll
see more of in 2011 is an increase in marketers promoting "application
specific SSDs" which solve a single problem or group of related problems
economically. It's tempting for SSD marketers to claim that their products work
well in a wide range of theoretical applications. While that might be true -
their products aren't necessarily affordable by everyone. For vendors who have
invested the time to qualify their products with named apps - the decision to
focus their marketing messages on niche user groups (while accepting the risks
implicit in ignoring other user segments outside those niches) should result in
faster sales ramps.
Being known as the #1 SSD vendor in a niche market
may be a better business than being the 183rd best known vendor in the wider
hidden segments in the enterprise for rackmount SSDs
||SSD patents and IP|
I predict we'll see more stories in
2011 about patent licensing and lawsuits related to SSDs.
technology markets - as the SSD market gets bigger in revenue - the opportunity
for companies to
from technology and IP licenses becomes a viable business prospect. In this
context "viable" means that a company which owns SSD related patents
is still likely to have change left over after instructing lawyers to launch
several lawsuits to protect its interests.
As SSDs get faster and
bigger in capacity - we're going to see some "IP stretch" as data
protection patents which were originally designed for
RAID systems and hard disk
arrays get redeployed into the SSD arena. In fact using RAID techniques to
protect SSD chip arrays isn't new - and some of these RAID chips patents go
back to the mid 1990s.
Patents related to data integrity, cache
management, parallelization, memory segmentation and memory virtualization are
some of the many topics waiting to be harvested by the IP legal ratpacks. That's
before encroaching onto silicon related IP such as better ways to manage MLC
What actually happened in the SSD market in 2011?
notes above were written before 2011 began.
For the month by month
summary read - 2011 - year of
the FIO IPO