|
|
|
|
| In popular fiction
about time travel it has often been said that the paradox you want to
seriously avoid is meeting yourself coming forwards while you are actually
traveling back. In writing about the computer market this is something
which happens to me a lot - as I look back on detailed technology and market
predictions
from the past and compare them to what happened. There's nothing too
mysterious about this swinging back and forth in content time. It's just
something which happens when I randomly dip into past pages I've written or
edited on the web over 2 decades to remind myself - what was all that about? One
of those prediction paradoxes occurred recently in connection with an SSD news
story in
August 2016 -
which intersected with a prediction article published in 2010 - which I had
illustrated at the time with this spoof press release. |
Stealth mode startup wakes
petabyte SSD appliance market
Editor:- October 17, 2016 - Exabyte SSD Appliance emerged from
stealth mode and today announced a $400 million series C funding round and
immediate availability of its new Paranoid S3B series - a 2U entry level Solid
State Backup appliance with 1PB (uncompressed) capacity. ...Latency is 10
microseconds (for accesses to awake blocks) and 20 milli-seconds (for data
accesses to blocks in sleep mode.) The scalable system can deliver 20PB of
uncompressed (and RAID protected) nearline storage in a 40U cabinet - which can
be realistically compressed to emulate 100PB of rotating hard disk storage using
less than 5kW of electric power. |
| Although some of the detailed
flash memory predictions
in that article 6 years ago were embarrassingly pessimistic compared to what we
expect to be seeing in the near term future - the purpose of that article -
this way to the
petabyte SSD - was to explore the user value propositions which would enable
solid state storage to become an attractive proposition to buy for high capacity
enterprise storage even if hard drive capacity cost less per physical bit. even
if enterprise hard drives are free And to make my point I used
the extreme
market boundary
condition that even if magnetic storage drives cost nothing to buy - they
would still be displaced by SSD based storage due to physical size, and
operating costs. The arguments I usedback then are accepted as
everyday
reasons to buy such systems nowadays - but I can confirm at the time of
publication they fired the imaginations of many movers and shakers in the
market. Since 2010 as I've written various stories about various SSD
systems companies I have from time to time wondered which company is going to
show such products first. Several years ago I thought a hot contender was
Skyera. But a few
weeks ago on a booth at FMS -
Nimbus Data was showing
boxes which looked very similar to this concept. Although the Nimbus boxes are
much faster than I predicted - because SSD power management has evolved to be
much faster at cold boot than the good-enough examples I used in my old article.
A critical ingredient of the spoof story and today's reality is
software. The
petabyte SSD was (I thought in 2010) an elegant wrap up and last bullet point in
a set of predicted
user value propositions for SSDs which I had been writing about since 2003.
Upto that point (in 2010) my technology predictions had been pretty
reliable at anticipating disruptive changes many years before any such products
or companies made them happen. But the set of user value propositions
wasn't as complete as I thought. Storage and CPU equivalence were to
be joined later by virtual memory equivalence. This happened because
the market experience of virtualizing storage and memory and leveraging
ideas from the early adopters in the enterprise
PCIe SSD market -
uncovered yet another big gap - tiered memory and trading latency for density
by adding more layers and a new core construct of persistent memory (with fast
memory boot which doesn't rely on preloading from magnetic storage). Thiese
permutations been playing out in the SSDs news pages in the past 2 -3 years but
the memory systems market still has a
ways to go
yet. what happened to big SSD ideas of 2016? Which
reminds me of the tentative title of this blog. It's traditional at
this time of the year for me to start collecting together and reflecting on the
big SSD ideas of the year. You can see recent examples of these "big
ideas" or annual round up blogs for various years here:-
2011,
2012,
2013,
2014,
2015.
In a long overdue break with past editorial tradition I'm going to
change the format for my big SSD ideas of 2016 article. And as part of this -
I've been asking people in the industry this question. From the
perspective of your company... what were the big SSD, storage and memory
architecture ideas which emerged and became clearer in 2016? Here
are some of the answers. |
"As 3D NAND gets designed
into all upcoming products we see everybody trying to manipulate and
characterize the flash at a low level in ways that they did not all need to do
before, partly because LDPC is required just to meet the specified endurance.
This is delaying roll-out for many because they either do not have the required
flash skills or they (other than our customers) do not have access to a tool for
automatically characterizing the flash and generating the LLR tables." Pearse Coyle, CEO - NVMdurance |
"The integration of the
PMEM/DAX driver into the Linux kernel allowing us to expose persistent memory as
a block device so it can tie into existing block and filesystem
infrastructure.... And also the NVMe over Fabrics standard and OS
support. Allowing us to present NVMe over RDMA to enable a new level of
performance for remote block storage." Stephen Bates,
Senior Technical Director - Microsemi |
"The major storage
manufacturers used the Flash Memory Summit (FMS) in August to make clear the
importance of Storage Class Memory (SCM).
Their choices tell us a lot about the limitations of their existing
NAND Flash technologies specifically in the area of endurance. It has been
known for a while that Intel and Micron have chosen 3D XPoint as their approach.
WDC stated that their approach will be ReRAM built into a 3-D NAND process
architectural frame.
What this means is that Intel, Micron and WDC have decided to jettison
a program/erase (P/E) mechanism, namely quantum mechanical tunneling, that has
been at the heart of NAND Flash since its inception in the late 1980's.
The improvements in endurance in going from 2-D to 3-D NAND have
clearly not been enough for Intel, Micron and WDC to remain with these
technologies for SCM.
Samsung on the other hand look like they wish to stay with a
tried-and-trusted P/E mechanism with their Z-NAND SCM offering. If Z-NAND is
indeed an SLC version of their V-NAND, then endurance is probably in the tens of
thousands. Schiltron technology is ideally suited for SCM in that we can reach
millions of cycles of endurance while remaining with the tunneling-based P/E
mechanism.
Furthering the concept of evolutionary development that has formed the
foundation of silicon technology since its beginnings, Schiltron uses existing
Fab tools and materials. We believe this will lead to the Storage Class Memory
Revolution." Andrew Walker,
Founder and CEO - Schiltron |
|
|
"Storage Class Memory...
As storage class memories are emerging, the memory hierarchy will be changed.
NOR-based NVDIMMs, such as 3D Super-NOR and 3D XPoint, will replace DRAM and SSD
at the same time. Also, software-based NVDIMM-P, such as HybriDIMM,
will come to the storage class memory market. Storage class memories mingles
fast-but-expensive volatile, and slow-but-inexpensive non-volatile memories
together. As a result, it will significantly boost system performance at low
cost and create huge market opportunities." Sang-Yun Lee,
President & CEO - BeSang |
"SSD capacities exceeding
HDD capacities are now shipping in volume this year in the enterprise." Greg Wong, President -
Forward Insights |
"The most significant
ideas we at Virtium see and are directing our resources toward center on taking
SSD security, reliability, durability, and manageability to a much higher level,
to meet the needs of increasingly connected industrial embedded environments.
These ideas are now being realized through drives self-encryption capabilities,
advanced remote-monitoring software, dramatically reduced power requirements,
and the development of SSDs specifically for the Industrial Internet of Things
(IIoT)." Scott Phillips,
VP Marketing, Virtium
Solid State Storage and Memory |
"From Mangstor's
perspective, we are heavily invested in NVMe over Fabric (NVMf) technology for
which our NX-Series Storage Arrays are based. With the ratification
of the NVMf specification this past June, flash devices such as SSDs and storage
arrays can now communicate over RDMA networks (such as RoCE or InfiniBand),
delivering the same high performance, low latency benefits as local attached
NVMe. Storage arrays based on NVMf, such as our award-winning NX6320,
avoid the lower level SCSI transport layer, which results in faster data
throughput and accelerated data access making NVMf solutions ideal for
compute-intensive HPC and database applications."
Scott Harlin,
Director of MarComms - Mangstor
|
"Nvme and soon 3D Xpoint
are affordable, solve low queue depth and write performance issues and will make
performance claims from the usual suspects irrelevant... Removing the interface
bottleneck was the giant step here." Guido
Meijers, System Engineer at Greenpower |
"I think definitely it is
NVMe. We have seen the demand has increased unbelievably this year. Most
customers told us that NVMe were their first choice of new systems design since
it combines ruggedness of traditional SATA and the high speed of PCIe."
Manager at a rugged SSD company |
"From our experience in
2016 we have seen 2 major things. First is the combination of storage
with the hypervisor (like our eEVOS). Without which customers would need to use
2 suppliers (storage and hypervisor) which would make support a nightmare
sometimes. The other is NV-DIMM techology, we have customers and
partners implementing it with our product. It provides best of both worlds - RAM
speed with extreme I/O and data still remains in case of server crash." Tvrtko Fritz,
CEO at euroNAS GmbH |
"For Hyperstone, the
biggest idea and industry trend that we pushed and participated in is the
implementation of page-based-FTL running on DRAM-less controller architectures.
This approach improves random write performance, increases endurance while
maintaining power-fail robustness at the same time. As this architecture also
reduces system cost it is also adopted in consumer markets.
We also see a significant adoption of USB 3.0 in industrial/embedded
markets. Certainly, the most hyped topics, at the FMS for instance, were 3D
NAND and NVMe but for both we do not yet see any sufficiently mature products or
any massive adoption in our markets" Susan Heidrich,
Sales & Marketing Manager - Hyperstone |
| interesting but... Not
everybody I spoke to was quite so focused on the immediate application of big
bang raw technologies. Some vendors - in
industrial and
military markets -
told me that their business focus in 2016 had been more biased towards
service oriented offerings. For example Limuel Yap VP
of Global Business Operations - V&G said that many
of V&G's efforts in 2016 have been in enhancing previously designed and
conceptualized products. And when I asked Camellia Chan
who is Managing Director at Flexxon
about this article she said - "Basically we are not the technology
leading company. Instead where we are focusing is on continuing to provide
legacy product to industrial, medical, automotive customers We support a lot of
EOL SSD products
worldwide. Most of our customers are facing the problems of discontinued
products and we are the ones who support them." did you notice
perhaps that something was missing? The above article was a
departure from the precedents established in previous years. Where -
you might ask - is your own set of suggestions Mr. SSDmouse editor? I'm
embarrassed to say I only had one. But I found a few other things to
add to my article which I hope will compensate for the unusual brevity of my
end of year list for SSD year 2016 and which (taken as a whole) portend
momentous things for 2017. Here's the link -
1 big
market lesson and 4 shining technology companies showing the way ahead for all
SSD and memory systems. | |
.. |
|
.... |
| Hmm... it looks like you're seriously
interested in SSDs. So please bookmark this page and come back again soon. |
.. |
|
| |
.... |
| SSD technology is the
mainstream. Demand is high. In many ways that's a good thing. But it's been a
mixed blessing because production of memories in the 2nd half of 2016 and during
most of 2017 did not keep pace with demand. |
| consequences
of the 2017 memory shortages | |
.. |
| A conversation I had (in
November 2016) with Charles Tsai, President - AccelStor provides a useful
example of how, nowadays, even the simplest type of SSD product plan has to be
aware of strategic considerations in a wide range of contexts which are sweeping
across the whole market. |
| a winter's tale
of SSD market influences | |
.. |
|
|
.. |
|
|
. |
| Retiring and retiering
enterprise DRAM was one of the big SSD ideas which took hold in the market in
2015. Over 20 companies have already announced products for this
market among which are Memory1, 3DXPoint etc But what are the
underlying reasons that will make it feasible for slower cheaper memory to
replace most of the future DRAM market without applications noticing? |
| latency loving
reasons for fading out DRAM | |
.... |
|
|
.... |
|
|
.. |
|
|
.. |
|
|
.. |
| Many of the important and
sometimes mysterious behavioral aspects of SSDs which predetermine their
application limitations and usable market roles can only be understood when you
look at how well the designer has dealt with managing the symmetries and
asymmetries which are implicit in the underlying technologies which are
contained within the SSD. |
| how fast can your SSD
run backwards? | |
.... |
|
|
.... |
|
|
.. |
| Can you trust market
reports and the handed down wisdom from analysts, bloggers and so-called
industry experts? heck no! - whatever gave you that silly idea?
|
| here's why | |
.. |
|
|
.. |
|
|
.... |
|
|
.... |
|
|
.. |
|
|
| .. |
| A couple of years ago - if
you were a big company wanting to get into the SSD market by an acquisition or
strategic investment then a budget somewhere between $500 million and $1 billion
would have seemed like plenty. |
| VCs in SSDs and storage | |
| .. |
|
|
| .. |
|
|
| .. |
| With hundreds of patents
already pending in this topic there's a high probability that the SSD vendor
won't give you the details. It's enough to get the general idea.
|
| Adaptive flash
R/W and DSP ECC IP in SSDs | |
| .. |
|
|
| .. |
|
|
| .. |
|
|
| .. |
|
|
| .. |
|
|
| .. |
Why
can't SSD's true believers agree on a single shared vision for the future of
solid state storage? |
| the SSD Heresies | |
| .. |
| There's one kind of market
research report which you won't find listed on the website of any storage market
report vendor - and that's a directory of all the other market research
companies they compete with! Here's my list - compiled from over 20 years of
past news stories - which includes all categories of market research
companies... |
| who's who in storage market
research? | |
| .. |
| If you spend a lot of your
time analyzing the performance characteristics and limitations of flash SSDs -
this article will help you to easily predict the characteristics of any new SSDs
you encounter - by leveraging the knowledge you already have. |
| flash SSD performance
characteristics and limitations | |
| .. |
| The memory chip count
ceiling around which the SSD controller IP is optimized - predetermines the
efficiency of achieving system-wide goals like cost, performance and
reliability. |
| size matters in
SSD controller architecture | |
| .. |
|
| Are you whiteboarding
alternative server based SSD / SCM / SDS architectures? It's messy keeping
track of those different options isn't it? Take a look at an easy to remember
hex based shorthand which can aptly describe any SSD accelerated server blade. |
| what's in a number? -
SSDserver rank | |
| .. |
|
|
| .. |
| A popular fad in selling
flash SSDs is life assurance and health care claims as in - my flash SSD
controller care scheme is 100x better (than all the rest). |
| razzle dazzling flash SSD
cell care | |
| .. |
|
| |
|