FMJ Storage, founded
in 2009, and headquartered in Woodinville, WA - designs "requirements
driven SSD solutions".
- mentions on StorageSearch.com, |
|What were the big
SSD ideas to learn and forget in 2015?|
who in SSD? - getting to know a Full Metal Jacket SSD maker
by Zsolt Kerekes,
editor - StorageSearch.com,
I spoke recently to FMJ - a company which (in
their own words and by their own choice) has - until recently - been flying "under
the radar" (of general market scrutiny).
The company's focus is
customers who have embedded applications in industrial, military and other
critical markets in which the reliability of every single SSD matters.
one of the 6 most fascinating conversations I've had in the past 18 months -
about the reliability and memory technology roadmap aspects of SSDs (in the
context of any market for SSDs and not just related to the
I learned some new things which it was good for me - and will be good for you to
Why is FMJ coming out now?
And what did I learn which
Read on - to find out.
Here are some
introductory bullet points about the company.
- FMJ designs SSDs which can include MLC or SLC.
- FMJ uses 2 main types of SSD controller:-
- one of which is their own design (which they don't want to talk too much
about - implemented by FPGA)
- and the other - a merchant market controller - to they have earned the
right to get special access from the original source - which enables them to
integrate their own firmware and access different parts of the controller for
various purposes (including power management) via normally restricted pins.
- FMJ's founders John Conklin and Dave Merry
have been in the flash SSD design business for over 20 years - and have been
involved in some of the best known companies in the past history of this market.
Despite that, however, FMJ has had to prove itself as a new company to
its suppliers - to get the access it needs to qualify and characterize
advanced emerging memory generations - and to get special access to modify
- FMJ has no outside investors and the company (at the time of writing this)
Why is FMJ coming out now?
- "We do not want the customer to ever send back parts (SSDs) that
don't work" said John
Conklin, COO - FMJ. (who among other things a decade earlier
consequences of SSD failure in the service life of industrial and equipment
applications can be a very high cost for customers - due to the lost
utility of those systems, or the cost of the unique missing data, and also
because of the high cost of sending engineers to possibly remote
That's because due to the physical
constraints of embedded systems in which the dimensions of space, weight and
power are all tightly bounded - the system design usually precludes the use
of the kind of
architectures seen in the enterprise.
So when a single SSD fails -
the whole system has failed too.
And in this context - a small amount
of data corruption - which might be recoverable by a request to retry - or
which can be fixed by an at rest healing process in an enterprise market
context - counts as a major systems failure in the industrial world. (That's
because Superman doesn't swing around and fly back again just because you
missed him flying past the window - the first time around. And BTW it's not
just birds, planes and the superheroes we're capturing data about.)
FMJ has already been designing and
shipping industrial SSDs into the market for several years - so I asked - why
is FMJ emerging from under the radar now?
(Having conceded that they
had done a good job of that - as they had entirely evaded my attention up to
the time they contacted me too.)
John Conklin told me that there were
2 main reasons:-
1 - they wanted to be sure that they had a bunch of
proven products and technologies (which they felt they were able to develop
outside the public gaze) and
2 - FMJ has found that even despite their
long experience in the embedded industrial market - they keep coming across new
applications which have never existed before. That suggests there are a lot of
other designs being done for new SSD applications by companies which aren't
the traditional companies or industries which are "known" to be SSD
prospects - even by experienced SSD sales and marketing people. The only way
to get into the awareness of this type of new customer - is to be more visible
on the web.
(This is similar to the business development problem within
the enterprise SSD market - which I have written about in my article about the
size of the Dark Matter
not just the same old industrial SSD
To demonstrated the concept of entirely new applications
and markets for industrial SSDs (some of which already have the potential to be
good markets in their own right) FMJ gave me 2 examples of applications which
come into this category.
At this point - I was able to surprise them -
by saying that I had designed a platform for one of these applications myself -
back in my pre-web career.
And I mentioned in outline one or two
examples of new markets for for industrial SSDs which I've read about -
which were new to FMJ.
All of which goes to prove the general point
- that the potential applications for embedded industrial SSDs cover a
much bigger area than ever before and the only way for SSD companies to learn
about those new applications and new customers - is to learn new skills of being
findable by potential customers on the web. - A necessary part of doing this is
that SSD companies have to less secretive - and say much more about what
they do - or don't do online - and why what they do is different. (This is
something which many SSD companies still feel uncomfortable about - because
it also educates their competitors.)
And another general point about
the industrial SSD market - is that it's becoming increasingly
The days in which a single SSD company can play the
lead role in markets as diverse as consumer, enterprise and industrial - may
be coming to an end.
industrial SSD customers are becoming more
focused and knowledgeable too.
And if a customer knows their
project is at the edge of what SSD technology can reliably do - the fact
that an SSD vendor is also a big player in unrelated markets - may count
for very little.
re FMJ - an afterword on meaning
Editor:- A few days
after publishing the above notes - I thought of another niggling little
question - which I put to John Conklin.
"I forgot to ask
what does FMJ stand for? - is it full metal jacket?"
- he replied.
Although that was trivial info compared to what I had
learned before - it made me feel better - as I like to think that even simple
things like product names might come from somewhere or have other meanings as
you can see in my article -
Speed and Strength Metaphors in SSD brands
|3D NAND flash challenges|
an industry roundtable discussion
|Editor:- February 6, 2014 - The best article
I've yet seen about the practical implications of increasing the adoption of 3D
NAND flash is this...|
At The Table: Commercial potential and production challenges for 3D NAND memory
technology - published by Semiconductor
Manufacturing and Design.
Among the many practical
considerations discussed in this article was the question of - "how is the
semi industry preparing for the transition to 3D memory?"
issue of scalability limits and market pacing - the article reveals that
vertical scalability currently appears feasible in roadmaps upto about 100 cell
But the rate of 2D shrinks in successive 3D designs
will slow down from the recent historic average of 20% per generation to 5% -
due to the problems of registration which accumulate as you add more layers.
|Naval Surface Warfare
Center (Navy) has approved FMJ Enhanced Security Technology PCMCIA Cards for the
Boeing F/A-18 Hornet a twin-engine supersonic, all-weather carrier-capable
multi-role combat jet, designed as both a fighter and attack aircraft. |
customer news (March 2015)|
|Editor:- February 18, 2016 - my inbox this
morning included a press release from Schneider Electric (not a name
anyone would associate with the SSD market) - but a nagging voice in my head
said - it might be worth opening. |
Sure enough a quick check on google
confirmed my guess. Isn't that the name of the company which acquired a
company I worked at a long time ago? (In 1980 I had an actual phy (non
in the programmable controller design group at a company called Square D.)
in the 1970s and early 80s industrial controller companies were at the
forefront of figuring out ways to design reliable digital systems using whatever
fancy new chips the semiconductor market was happy to toss at the world. (That
was before the chip companies got seduced by the PC market - after which they
cared even less about industrial applications.)
From what I can
remember of those days (and wavering the warranty on my mushy organic
carbon memory ECC) there seemed to be many examples of devices and circuit
design techniques which worked perfectly OK in a computer - but which would
die a swift death in a harsh industrial environment. And unlike the military
market (which had similar environmental constraints) industrial equipment
designers had to make things work within much smaller budgets.
- back to the present.
Everspin's MRAM is
being included in a new generation of
controllers. I find this interesting as it provides a vote of
confidence in the rugged operational integrity of this newly viable memory
13 years of progress on
"MRAM will soon replace flash",
|how safe are your
assumptions about SLC? |
|Editor:- March 14, 2014 -
is regarded as the "gold standard" in
nand flash memory today
when it comes to
Or maybe it would be more accurate to say - "SLC is the depleted
uranium standard" when it comes to choosing ingredients for hardening the
SSD data integrity
So you can imagine my surprise- when in a recent
conversation about the reliability aspects of SSDs - I was told about some
unique and proprietary "brutal and awkward test patterns" - which
had uncovered design flaws in a new type of SLC memory while it was being
characterized for use in SSDs.
This indicated that SSDs designed
using that memory in some applications could be killed in as little as 3 to 9
months of use.
This design vulnerability never showed up at all
in the "standard"
SSD controller test
patterns which are used throughout the industry. And their application wasn't
for an SSD accelerator - but for a regular speed SSD.
customer point of view - if you want an embedded SSD which you can rely on -
it's nice to know that some people still design SSDs the old fashioned way - and
test every assumption along the way.
That was just one of many new
things I learned talking to Dave Merry
Conklin co-founders of a new SSD company called FMJ Storage - which has -
for the past several years been operating profitably while under the general
|FMJ - re SLC, MLC and 3D|
|Editor:- March 17, 2014 - Trying to make sense
of the changing patterns of memory usage in SSDs and guessing which way
things are likely to go - is something which all SSD designers and specifiers
think about. |
But because of the unusually long timescales for
compatible product availability in the embedded industrial markets - which
start at 7 years typical in-service life - added to the extended temperature
operating range (compared to other markets) means that the the game of "guess
what flash memory will still be availabile roadmap" is different in
I was fortunate to get a personal view on the long term SLC
and MLC questions - as they relate to the
- from one of the most experienced people in this industry - Dave Merry -
whose SSD career (as VP of Engineering, CTO or cofounder) has spanned some of
SSD history's best known industrial SSD companies - SiliconTech (which later
became SimpleTech and then
Modular and SiliconSystems.
It was in a recent conversation with Dave at his present company -
FMJ Storage that I got to
ask these questions.
Zsolt - How long do you
think will SLC continue to be available?
Micron has said it will
continue to provide SLC for some government projects for 15 to 20 years.
- I didn't know that.
Zsolt - Wer're seeing a
lot of complicated
controller technologies being used to make MLC usable and improve
How do you see MLC going?
Dave - We have to keep abreast of all
technologies - and have had early access to 3D nand. Our
characterization suggests that the endurance of 3D nand is 3x to 4x
better than MLC (at similar line widths). So that makes 3D a good place to look
at for a long term MLC industrial roadmap. But it's too early to know if the
3D layout and architecture itself might have its own (currently unknown) new
failure mechanisms or R/W disturbance sensitivities - which would require new