 |
leading the way to the
new storage frontier | |
..... |
SSD news - January
15 - 31, 2013Companies mentioned on this page in stories include:-
BiTMICRO,
Diablo Technologies,
Foremay,
Fusion-io,
GridIron Systems,
iSuppli,
Micron,
Nimbus Data Systems,
OCZ,
Proximal Data,
RunCore,
Seagate,
Skyera,
STEC,
Violin,
Virident,
Virtium,
WhipTail |
. |
Virtium ships 512GB industrial Slim SATA SSDs
Editor:-
January 30, 2013 - Virtium
today
announced
it's shipping true industrial
512GB SSDs in the Slim SATA (MO-297) form factor - which is just 15% of
the physical volume of a standard
2.5" drive.
Editor's
comments:- it's interesting what you can do with numbers - and I admit I hadn't
thought of that 15% volume comparison thing even though I've been writing about
tiny (and
not so tiny) SSDs for a
while.
But seriously the comparison breaks down if you need more
capacity - because if you need 2TB (say) then 4x MO-297 is 60% of a 2.5"
drive which reduces the strength of the argument. On the other hand - the common
applications for these smaller embedded drives - networking and ATCA blade
systems - according to Gary Drossel
at Virtium - don't need such big storage capacities (yet).
RunCore is 1st to announce BiTMICRO OnBoard
Editor:-
January 29, 2013 - today
announced
that RunCore
will use BiTMICRO's
Talino controllers in its new
Kylin III MAX
family (fast PCIe SSDs).
Diablo sets up compatibility team for new SSD interface
Editor:- January 29, 2013 - Diablo Technologies
today
announced
it has set a compatibility advisory team for its new SSD interface - which
the company is apparently positioning as a faster alternative to
PCIe SSDs.
"As
we prepare to launch our line of Memory Channel Storage products that enable
next-generation enterprise server and storage system designs, we have set our
sights on unprecedented levels of performance for current and future
applications To that end, we have
assembled
a group of top industry innovators to help refine the development of our
revolutionary NAND-flash system solutions..." said Diablo's CTO - Maher Amer.
See
also:- how
will Memory Channel SSDs impact PCIe SSDs?
Seagate turns to Virident for big SSD architecture
Editor:-
January 28, 2013 - Seagate
today
announced
it has made a strategic equity investment in Virident as part
of a new collaboration agreement which includes remarketing Virident's
PCIe SSDs and working
together to design new SSDs
for the enterprise market.
"Seagate is thrilled to team with
Virident, a technology leader in one of the fastest growing markets in
enterprise and cloud computing," said Gary Gentry,
senior VP and GM, Solid State Drives at Seagate. "Together, we are working
to develop the next-generation hardware and software solutions in the
PCIe space."
Editor's
comments:- it was obvious a year ago that Seagate's earlier marriage of SSD
IP convenience with LSI
wasn't going to last long or remain monogamous - as LSI and Seagate would be
competing for the same oem design slots in the enterprise market and
furthermore LSI's small architecture
SandForce controller
isn't efficient for
multi-petabyte
scale fast SSD installations. (And in the
consumer market
LSI didn't have the
adaptive
controller IP - which is what led to Seagate's stake in
DensBits)
Virident
- a top 10 SSD company
- has a roadmap
scalable big
architecture enterprise SSD controller and drive family which has been
developed in a cleanroom environment - where all the critical IP has been
devleoped by the company.
The obvious gap in the Seagate / Virident
product line is a 2.5"
removable PCIe SSD (to compete with
Micron) - and it's a no
brainer to see that Seagate's experience with this form factor - coupled with
Virident's SSD design skills could quickly result in viable products for this
new market - which will replace upto
25% of the
projected market for fast
SAS drives.
Proximal Data - case study
Editor:- January 28, 2013
- Proximal Data
today
announced
details of a
new
case study (pdf) re the use of its
SSD ASAP software (AutoCache ) to
trim 30 hours off the monthly SAS analytics report for a financial customer
which used to take 36.5 hours.
Editor:- when
SSD software companies
start talking about real customers - the
acquisition
press releases follow not long behind based on
recent SSD history.
STEC launches 2TB SAS SSD
Editor:- January 28, 2013 -
One of the oddest linking ideas I've ever seen in an SSD news story appeared
today in a
press
release from STEC
which suggests that anyone should care that the company is the first in the
market to launch both a 2TB PCIe SSD ($9,425) and a 2TB SAS SSD ($7,995).
This
marketing communication undervalues the true achievement of a 2TB
SAS drive (assuming it
fits in a standard height - unlike a previous model).
Also nearly
lost in this mixed up marketing communication is the idea that STEC is offering
"Unlimited Writes" on some variations of these products. (Which
endurance
goal is possible using a variety of different techniques - at the extremes
being slower performance, using more expensive flash or RAM caching - I haven't
asked which of these applies.)
OCZ's new VP of Global Sales
Editor:- January 23,
2013 - OCZ today
announced
the appointment of Wayne B.
Eisenberg as Senior VP of Global Sales.
Eisenberg's previous
experience with SSD customers and technologies includes 16 years at
SMART Storage Systems
iSuppli says SSD shipments in 2016 will be 6x 2012 level
Editor:-
January 23, 2013 - iSuppli
today
predicted
that worldwide SSD shipments this year will rise to 83 million units this year,
up from 39 million in 2012.
iSuppli also said it anticipates that in
2016 - SSD shipment volume could be 239 million units - equivalent to 40% the
size of the hard drive
market.
another new VP at Skyera
Editor:- January 23, 2013 -
Skyera is
still fleshing out its VP team with the
announcement
today of new VP of NAS technologies - Amit Bothra.
See
also:- SSD
efficiencies,
adaptive R/W,
rackmount SSDs
Micron's hot pluggable SATA SSD
Editor:- January 23,
2013 - Another day, another
SATA SSD - this time
from Micron -
a top 15 SSD company
- who recently reported a significant transition in its memory to SSD company
roadmap - so stop yawning at the back there! - and whose new hot pluggable,
regular cached,
MLC enterprise
P400m
SSD (rated at 10 drive fills / day for 5 years) is said to offer the same
endurance as
SLC at under half the price - if I have read their chart correctly.
UBER is <1
sector per 1016 bits read. Performance isn't something to write home
about - but is adequate for a wide range of fast-enough array applications. ...datasheet
(pdf)
See also:-
the new fad in selling
flash SSDs... life assurance and health care claims
Nimbus ships petabyte SSDs / month
Editor:- January
22, 2013 -
Nimbus Data Systems
today announced
it has been shipping at the rate of over 1
petabyte of SSD
storage / month.
Violin acquires GridIron
Editor:- January 21, 2013 -
Violin today
announced
it has acquired GridIron
Systems.
Editor's comments:- in
October 2012 I
listed GridIron as 1 of the 3 main contenders to
Fusion-io in the
enterprise SSD software
stakes -with the qualifying comment...
"GridIron - probably has
the most sophisticated SSD
ASAP software in the industry. But it's a shame it has been tied (until
recently) to their hardware - an SSD HDD hybrid box."
Today's
announcement - which adds to the growing list of
notable SSD
acquisitions in the modern era of the SSD market - will enable Violin to
strengthen its already established authority in the enterprise SSD rack market.
Can you guess WhipTail's revenue?
Editor:- January
17, 2013 - Sometimes you can estimate an SSD company's revenue even when
they're a private company and coy about the numbers.
For example -
WhipTail's
announcement
that its revenue grew 300% in
2012 - coupled
with the statement that it has shipped a total of over 3.7 petabytes of SSD
storage - and assuming the
average market price
per terabyte for fast enterprise rackmount SSDs - is all the information you
need to make a reasonable estimate.
It's also interesting to compare
those petabyte
shipment numbers with disclosures from other enterprise SSD companies in
earlier years.
Don't send me your answers - because there's no prize
for the right answer. Just the sense of satisfaction that you can do it. (Hint -
this method of calculation gives an answer which is similar in order of
magnitude size to the company's recent funding round - which if you're familiar
with the cash flows you need to support that growth rate provides a satisfying
sanity check.) Now - the only question is - is WhipTail's revenue growth rate
sustainable?
Market
demand curves signal yes.
Competitive pressure
factors signal no.
But even if you apply a dampening factor -
and assume only 200% revenue growth in 2013 - you're still looking at another
enterprise SSD company which you can't ignore. (Not that you would anyway - as
the company ranks #6 in the
top SSD companies
searched for by our readers in Q4 2012.)
Fusion-io positions ioScale for new SSD Dynasties
Editor:-
January 16, 2013 - Fusion-io
has released a new PCIe
SSD called the ioScale
(3.2TB on a single half length PCIe slot) which is aimed at technically savvy
customers who have the potential to use thousands of cards in their
installations in new
dynasty enterprise SSD apps.
Pricing is under
$3,900 / TB and the minimum order quantity is 100 units.
Editor's
comments:- When you first look at this product - you might be tempted to
think - So what? - isn't it very similar in capability to other products which
FIO (and others) have shipped already?
In one way you'd be right. The
ioScale's hardware design is based on FIO's experience in making low cost PCIe
SSDs for the workstation market - which is as close to
consumer market
price pressure as FIO gets at the present time.
But the ioScale is
aimed at a special class of enterprise super users - whose apps and companies I
call:- new dynasty
and dark matter
respectively.
Rick
White CMO
Fusion-io told me
that when they did market research into the kinds of customers who were already
using their SSDs they discovered the big enterprise SSD customers could be
segmented into 2 groups which superficially had similar performance needs - but
were very different in the ways in which they valued issues such as:-
- compatibility with traditional software apps,
- how they handle reliability,
- how often they refresh and replace their infrastructure.
- how they assess the cost / benefit of features within SSDs
The
traditional enterprise customers have the profiles which everyone in the
industry knows about and aims their products at - but the new type of enterprise
customers have needs which are only starting to clarify - and for this latter
type of customer - SSDs are a strategic business enabler - because they can
convert efficiencies in raw computing technology into real competitive
advantage.
Fusion-io is one of the few companies in the world which
already has a set of these latter cloud / data factory economy customers
who each have already got thousands of high performance PCIe SSDs - and who
have the ability to scale up substantially if their requirements are met and the
SSD enabled economy grows in the directions expected.
Rick told me that
these customers do want
scalable SSD
performance, and low cost - but they don't need many of the bundled frills
which are deemed to be necessary for traditional enterprise SSD customers
When
legacy apps report faulty drives they change the drive or the rack. When
uber new dynasty SSD
users report faults - they route around them. Then when the time comes to
upgrade the CPU and storage capacity per square foot of that region in the
datacenter - the whole lot is forklifted out and replaced - faulty and unfaulty
racks - makes no difference.
Also - in these apps - hot pluggable
drives are a frill which is simply not worth paying for.
The dark
matter SSD customers - at which the ioScale is aimed - also know much more about
the technical limitations of their infrastructure - and have the technical
expertise to change things to suit them better - if they think it's worthwhile.
So - for example - the ability to dive into SSD APIs and change their apps
code to get speedups or other new functionality - is something they will do -
whereas traditional enterprise customers prefer all new hardware to work with
pre-existing software in a tweak-free environmoent.
During my
conversation with Rick White - I referred back to the
ION
software (which FIO launched in
August 2012 -
and which enables users to convert a standard server and a bunch of
PCIe SSDs into a
traditional SAN compatible
rackmount SSD).
My assessment of that product shared with readers at the time -
was that if it satisfied the needs of a small number of super users - who could
each buy maybe hundreds or thousands of such systems - that made it worthwhile
for FIO to bundle the concept and launch it. I thought the analysis I had seen
in other places - which compared it to traditional rack SSDs was completely
missing the point.
Rick confirned my analysis was closer to the mark -
and many times in our discussion we returned to the problems in the SSD market
caused by faulty and incomplete
market research and
mistaken understandings of what the real issues in the market were.
My
way of summarizing this is - that if you ask a bunch of people who go to a
trade show - what do you think about SSDs? - you're going to get a different
result to when you talk to people who are already deeply engaged in the SSD
market, have already done a lot of SSD projects and who spend nearly all their
waking hours thinking about what more can they do if they had even better SSDs?
It's
not that the traditional market research gives you the wrong answers - it's more
that - if you're not in the right place in the SSD market then you don't
understand enough to pose the right questions - and you probably don't have
access to the people who will ultimately decide the answers.
Fusion-io
isn't the only SSD company who is getting value business insights by
researching its strategic customers.
I reported last year that
SanDisk had adapted its
approach to enterprise customers by deciding to support competing hardware
with its
FlashSoft software.
And there are many more examples I could mention if I had the time.
Foremay ships 2TB 2.5" SSDs for industrial market
Editor:-
January 15, 2013 - Foremay
today
announced
immediate availability of 2TB
industrial
2.5"
SATA SSDs with
standard 9.5mm thickness.
Editor's comments:- In its press
release for this product Foremay claims to be the first company to do this. But
this is one of those situations where I think being "first" tells us
more about market conditions (where things stand in the
HDD replacement
part of the SSD market) than about the technological supremacy - or
otherwise - of any particular SSD oem.
It's been technically feasible
to make 2TB 2.5" SATA SSDs for the past 2 years. The only reason you
haven't seen them flooding into the market is that such products would have been
unattractive before to both SSD oems and to SSD buyers.
For the SSD oem
- the same bunch of memory chips used to make a 2TB 2.5" SATA drive - have
been much more profitably deployed in faster SSD modules such as
PCIe SSDs or
SAS SSDs.
Meanwhile
for SSD buyers and specifiers - 2TB 2.5" SSDs would have been very
expensive -
compared to the alternatives - while delivering no performance benefits (due to
the slowness of the SATA interface) which means that 2x 1TB SATA drives are
faster in a storage system then 1x 2TB drive.
In today's market,
however, the cost differences between flash SSDs and hard drives have shrunk to
the extent that for industrial equipment designers who look at the cost of
reliability in a
7 year industrial product operating life timeframe - the alternative of using
1 factory fitted SSD compared to the probability of using 2-3 HDDs in the same
slot (taking into account the MTBF distribution over the system population)
makes high capacity SATA SSDs attractive.
This class of SSDs will
also extend the market life of equipment designs which originally used
HDD interface standards as a way of virtualizing and protecting against
generational changes in raw flash
memory - even if they never used hard drives. The extra storage capacity
enables equipment designers to integrate more features and software. | |
. |
|

| |
..... |
|
..... |
top 100 SSD articles Can you
trust SSD market data? Adaptive R/W and
DSP in flash SSD IP the Silo classification for
all enterprise SSDs Symmetries and
consequences in SSD design Efficiency - making the
same SSD - with less flash design factors which
shape flash SSD performance
|
. |
 |
. |
"In some ways, blocks
lost due to media corruption present a problem similar to recovering deleted
files. If it is detected quickly enough, user analysis can be done on the
cyclical journal file, and this might help determine the previous state of the
file system metadata. Information about the previous state can then be used to
create a replacement for that block, effectively restoring a file." |
Why
CRCs are important - blog by Thom Denholm Datalight (January
2013) | | |
. |
 |
. |
"Thanks
for the offer, but... we don't want to deploy any new hard drive arrays.
Not even if you're giving them to us free!" |
This classic article described the pivotal
future storage market climate in which enterprise users will cease to regard
hard drive arrays attractive or usable - even if the cost of buying a new hard
drive array drops away to ZERO! -
this way to the
petabyte SSD | | |
. |
 |
. |
If an SSD company is worth
significantly more or less (or about the same) on a Friday than it was on the
Monday... |
and the problems of trying to unstick a sticky SSD platform
Editor:-
January 30, 2013 -
Fusion-io's
results will be announced when the markets close today, and because of FIO's
position in the SSD market - it seems like many people just hold their breath
and wait to see what happens.
I wrote about this tense waiting
period for investors, customers and (even) competitors in an article 3 months
ago called -
SSD
investments, x8 flash, and ETs - which may calm you down if you're feeling
anxious.
...Later:- January 31, 2013 - the
results
reported were:- revenue of $120 million in the recent quarter, up 43%
from the year ago period.
For any other company that would be a great
result - but FIO had to explain to stakeholders why its revenue growth
apparently slowed down. More details in their press release.
Even
before looking at what FIO said I wasn't surprised to see a small slowdown in
growth. With so many different ways to do the same type of acceleration now in
the market - and so many complex new
SSD software
options available - I suspect it's taking even dedicated FIO customers a bit
longer to evaluate and optimize the exact configuration of the solutions they
need - even when they have already bought into the idea that they want more of
something that's nearly the same.
Customer pragmatism comes into it
too. The SSD platform decisions being made now - will have to make sense in 5
years time too.
...Later still:- February 1, 2013 - a reader
asked me to expand on the above point - particularly given his concern that FIO
is suggesting that revenue may be even lower in the next quarter.
Please
bear in mind that this is my own analysis and speculation of what may be
happening in the market right now. Most of the text below is cut and paste from
an email I sent to a reader shortly before putting it on this webpage.
re
- what could cause big enterprise SSD users (in
new dynasty apps)
to slow down their uptake of SSDs for a while - given that
analysts like me
have projected that the total adoption in this market will grow a lot this
year?
The dynamics of big SSD customers go like this...
They
like SSDs.
They've seen details of new upcoming products or memory
technologies which look cheaper than what they're buying but maybe the
software for the new stuff isn't what they want yet.
If they're a big
Fusion-io customer they obviously like FIO but if the customer isn't careful
how they fit the FIO stuff into their company framework they know they'll be
locked into FIO just as much as previous generations were locked into IBM360 or
VMS or Solaris so they're looking at all the other things in the market -
including other suppliers and ways of doing things - and trying to figure out
- is there a way to use more of this FIO stuff easily without getting locked
in?
And how about the other
SSD silos in the
enterprise where FIO isn't a supplier - but somehow it all has to play
together?
There are many variables to optimize - and it's naiive to
pretend they can all be kept separate and managed independently forever.
THE
SSD SOFTWARE PLATFORM will become the way that users will interpret all SSD
purchases in future. I think about 5 different companies have a viable
prospect of owning that enterprise front end and being important gateways in
the enterprise SSD world. I've talked to most of them and to other companies
who want to be the SSD platform.
Prospects for users are scary as they realize SSDs are not a short
term fix but will become the biggest place they put their future server budgets
in. It means that users will wait and try and figure things out. But if they
wait too long their infrastructure slows down. gets less competitive.
The odds are if you've already got FIO (or some other vendor's key
SSD tool which works) you are already locked in but just don't want to admit it.
Are you safer using FIO through a 3rd party -
SanDisk /
FlashSoft platform?
yeah maybe - if that's possible - but then you lose the performance - you
could get from embedding Fusion-io's APIs.
And BTW SanDisk has
only been in the software business for a handful or quarters.
ditto comments re - OCZ's
VXL
Customer navel gazing and
SSD software lock-in
concerns are real and serious. But not easily solvable.
Once you start using SSDs you can't stop using more.
The future SSD purchases will have to be made (with someone). Some problems
have no tidy solution. Figuring out the best way to architect SSD datacenters
and creating internal standards takes time. But there isn't any time.
Another simpler answer - given in FIO's own conference call - but
which I hadn't heard till after writing this piece - is that if the customers
are getting cleverer at using SSDs - what CEO, David Flynn referred to
as their increased efficiency
- then in the short term they can do the same workload they planned earlier with
less SSDs - for now. But it's reasonable to assume that will simply lower the
viability threshold for using SSDs in other new functions too - because the SSD
cost per added business customer value is better. Which means even more SSD
demand from new apps down the road.
Meanwhile users who have stuck
rigidly with legacy SSD architectures - think they don't have to worry so much
about these issues - because the SSD fits into their pre-existing schema for
doing things. But even the legacy SSD solutions have started to add personality
traits. And even legacy architecture SSDs will have to fit into your future SSD
software platform.
Legacy apps users can delay for now making these
decisions and still avoid getting too stuck into any single supplier - but they
will have to go through the same tortuous company-wide SSD needs analysis
themselves too - in the next year or so too - in order to get the best value for
what they're doing. | | |
. |
 | |