| leading the way to the
new storage frontier||...|
trust SSD market data?
sudden power loss
flash story... could it have been simplified?
hidden preference segments in enterprise flash
of enterprise SSD companies have no good reasons to survive
"We currently forecast that ASP per GB for enterprise SSD will decline
to around $0.37 over our forecast period. And global enterprise storage units
will grow from 10.5 million units sold in 2014 to over 30 million devices ."|
(analysis of SanDisk) June 2015|
market research resources
|Altera launches adaptive
DSP controller for PCIe SSD market and Intel's acquisition of Altera from an
|Editor:- June 23, 2015 - Altera today
availability of a new flash controller reference design for the
NVMe PCIe SSD market
which uses adaptive
writes and DSP ECC. The
10 SoC (pdf) which includes among other things an integrated dual-core ARM
processor uses flash IP from
Mobiveil and NAND
optimization software from
simplify the design of gen 3 PCIe SSDs having 7x better endurance than
classical non adaptive designs.|
Editor's comments:- Since the
market criticality of adaptive DSP flash controller techniques for enterprise
SSDs started to emerge in 2011 and then clarified in a big way in 2012 - it
has become an essential capability for most product lines. This standard
product from Altera fills a much needed gap in their offerings.
how will Intel's acquisition of Altera affect SSD market?
this month:- Intel
it had agreed to acquire Altera
for $16.7 billion.
I don't think it will change any of the
fundamental technology directions in the SSD market. But I did discuss it with
some readers who asked me about related issues. Here are some extracts from
what I said in various emails.
The Altera acquisition makes perfect
business sense - because Intel had lost out on many big markets (such as mobile
phones etc) due to its unwillingness to design custom solutions for specific
Intel's inability to make that kind of business work (where the
customer leads the architecture) was demonstrated back in the late 1980s with
their ASIC business which was based on gate array technology which they
obtained from IBM in return for rights for IBM to design custom X86 processors.
Unfortunately the IBM ASIC technology was unwieldy and less well
supported by low cost EDA tools than many of the competitive offerings from
pure play gate array and standard cell companies. So the ASIC technology was
unattractive outside a small core customer base - and soon fizzled out. -
But IBM got to keep the more valuable rights to the X86.
other market lessons where Intel experimented but got burned (such as the
digital watch and DRAM) that lesson remained imprinted in future Intel
management culture - that there are some markets which Intel should avoid
participtaing in with market specific silicon products:-
- those which have the potential to be commodities (like memory) and
provides a way of market customization via a standard product.
- those which require high degrees of customization and custom architecture
for one specific product or customer and where Intel's architecture and
legacy software ecosystems are not the central themes of the product.
FPGAs from Altera and other companies are widely used within enterprise SSD
systems and also within low to medium volume embedded SSD drives too.
this acquisition - which gives Intel a market leading reprogrammable
controller platform - will enable marketers and technologists in Intel to
stick to the comfortable concept of predictable semiconductor geometry based
roadmaps - while also having an engagement within the SSD market and visibility
of trends which goes much wider than their previous product lines enabled.
Flash Memory Summit - diary date|
|Editor:- June 15, 2015 - Conference Concepts today
announced registration is live for the 10th annual Flash Memory Summit, the
worlds largest conference dedicated to
flash memory and its
Memory Summit 2015 showcases the latest in flash memory design with over 10
simultaneous tracks and 14 keynotes speakers.
The 2015 keynote speaker
line up features the industrys flash memory leaders including Samsung, SanDisk,
PMC, Toyota, Micron, Oracle, Tegile, SK Hynix, Toshiba, Kaminario, and Diablo
Vendors interested in exhibiting, please contact Alan Land at
1.760.212.5718 or Sales@FlashMemorySummit.com.
Editor's comments:- I don't go to physical trade shows
because due to the efficiencies of electronic communication and after
20 years of publishing
on the web (and with the help of my readers) every day here at
StorageSearch.com feels like an SSD trade show.
Nevertheless if that
were not the case - and if I did have to pick one physical event at which
to get immersed in current thinking about the future of enterprise SSDs - it
would undoubtedly be Flash Memory
I know that many leading thinkers in the industry will be
polishing up all their best ideas to pack into their presentations at the
show. And it's a rich resource for anyone trying to get the lie of the land in
a short space of time.
But even if you don't have the inclination to
drag yourself to Santa Clara in August - the FMS web site - which includes
archived content from past years - is a rich resource for tapping into a wide
range of SSD oriented memory thinking.
| hold up
capacitors in 2.5" MIL SSDs|
do you really need them?
Editor:- I've been looking at different aspects
of power hold up schemes in mission critical non volatile memory systems for
over 30 years. |
But every time I revisit this vast topic and compare
fresh examples from the market - I learn something a little bit new.
new blog -
to three seconds - demonstrates the extreme range of hold up times
now in the market inside leading edge 2.5" military flash SSDs.
|90% of the enterprise SSD
companies which you know have no good reasons to survive
|In one of the most highly read
on StorageSearch.com in
recent years - I looked at drivers, mechanisms and routes towards
consolidation in the enterprise SSD systems market along with some other
outrageous and dangerous ideas. The conclusion?|
the enterprise SSD companies which you know have no good reasons to survive."
publication - I discussed these ideas with various readers for about 3
months and since publication you won't be surprised when I tell you it has
been at the core of many conversations since. ...read
Junes of yore in
|1 year ago - June 2014 -
announced it would acquire
a transaction valued at approximately $1.1 billion.
3 years ago - June 2012 -
a guide on the market readiness and likely impact of
flash care management & DSP IP in SSDs
4 years ago -
- NVSL demonstrated the world's
first PCIe SSD accelerator based on PCM (phase-change memory).
years ago - June
2010 - Anobit
announced it was sampling SSDs based on its patented Memory Signal Processing
technology which could deliver 20x improvement in operational life for
MLC SSDs in high IOPS server environments.
13 years ago - June 2002 -
raised the capacity of its industrial DiskOnChip SSD to 1 gigabyte.
years ago - June
2001 - Adtron
shipped the world's highest capacity
3.5" SSD. The
S35PC had 14GB capacity, could operate in rugged environments and cost
|If you could only get 3
messages to the most important people in your industry to help them understand
your view of the market what would they be?
|NexGen decouples from Fusion-io accelerator
Editor:- June 30, 2015 - As previously signaled - NexGen Storage has
decoupled itself from relying on SanDisk's PCIe SSD
product line in its hybrid storage arrays with the
today that NexGen has introduced NVMe readiness as an update in its software
services. This paves the way for expanding the systems product line with a
wider range of
3rd party internal SSD accelerators with different
Swissbit samples new range of rugged mSATA (MO-300A) SSDs
June 29, 2015 - It's been over 4 years since I last heard from Swissbit - but this
week the company
it is sampling a new range of mSATA
Series with random R/W performance of up to 75,000 IOPS (4KB). These
rugged SSDs (shock / vibration rating of 2000G / 20G as per MIL-STD810) use 15nm
MLC from Toshiba and are
available with capacities from 8GB to 480GB.
what does Tegile's customer survey tell us?
June 25, 2015 - A
published survey conducted among customers of Tegile reveals some
interesting insights into the demographics of Tegile's business but also - just
as interesting - provides a spectrum of weighted answers about why people bought
enterprise flash arrays and the perceived benefits.
Among the many
results - the most interesting for me were:-
- "36% plan to use their Tegile storage to accelerate the
development of new products and services. Customers can create read/write clones
of production databases. This enables them to get new applications into
production faster without consuming a lot of storage space."
comments:- that's a classic enterprise SSD advantage related to a pain point
which I was discussing with users over 10 years ago.
don't have the performance freedom which SSDs deliver - but who struggle even
to keep their legacy platforms running sluggishly - know there must be better
things they can do with their raw business intelligence - but are too scared to
interact with the production data. And designing new systems based on sampling -
doesn't give the full picture.
survey findings which I've commented on above - probably apply to any
leading AFA vendor and not just Tegile.
- Tegile says "51% of customers expect to see an ROI in 12
months or less."
Editor's comments:- that's confirmation of
something I said in my article -
year of the enterprise
"what's driving this confidence is that their
customers have done the pilots- they've done the product tweaks - the biggest
customers have finished their cautious rollouts - and they're coming back asking
for more than more. The user mood is changing from - can I afford to use SSDs?
- to a realization that - I can't afford not to use SSDs." (October 2011)
Tegile's survey confirms is that the same advantages which were first
experienced by early adopters do indeed trickle down and deliver similar
impacts to mainstream users (if the products are
a way which is attractive enough to tempt new customers to experiment.
Here are some interesting
results which are specific to Tegile's business.
- Re the importance of offering "unified" connectivity (FC+IP) - "over
half (53%) were using more than one storage protocol with their Tegile
Editor's comments:- that's a higher proportion than I had
assumed. Which also is consistent with the broad spectrum of traditional storage
suppliers that Tegile has been displacing (another aspect shown in the survey).
the article - Why do people use Tegile Flash Storage?
- Re customer satisfaction? - 92% of responders said that they'd
Editor's comments:- that's a good story for a company
whose business model has been so reliant on
external funding to
sustain its growth.
But how sticky is brand loyalty in the enterprise
flash market? Especially when we're entering a period where I predict that over
of enterprise SSD brands will disappear?
Let's just say that high
customer satisfaction is an excellent achievement but that a customer who has
switched once because they saw a good reason to do so - is a customer who could
easily switch again. While Tegile has some sticky service and software
solutions in its product delivery - don't be beguiled by statements like the
above - if and when you consider the IPO.
And - on the
of extrapolating inferences from surveys - 96 users - the customers who
took part in the survey - aren't the whole market.
SSD market slowing down?
Editor:- June 22, 2015 - In
a new observation on the state of the SSD market -
SSD Insights Q2/15:
Slowing Down - Gregory
Wong, President, Forward Insights
said - "The weak PC market and tepid datacenter demand affected shipments
of SATA SSDs in Q1/15.
This was offset by strong shipments of
SAS SSDs and SSDs into
the channel which benefited from aggressive pricing, particularly in Asia."
bath tub curve is not the most useful way of thinking about PCIe SSD
failures - according to a large scale study within Facebook
June 15, 2015 - A recently published research study -
Study of Flash Memory Failures in the Field (pdf) - which analyzed
failure rates of PCIe
SSDs used in Facebook's infrastructure over a 4 year period - yields some
very useful insights into the user experience of large populations of
Among the many findings:-
- Read disturbance errors - seem to very well managed in the enterprise SSDs
The authors said they "did not observe a statistically
significant difference in the failure rate between SSDs that have read the
most amount of data versus those that have read the least amount of data."
- Higher operational temperatures mostly led to increased failure rates,
but the effect was more pronounced for SSDs which didn't use aggressive data
throttling techniques - which could prevent runaway temperatures due to
throttling back their write performance.
- More data written by the hosts to the SSDs over time - mostly resulted in
more failures - but the authors noted that in some of the platforms studied -
more data written resulted in lower failure rates.
attributed to the fact some SSD software implementations work better at
reducing write amplification when they are exposed to more workload patterns.
- Unlike the classic bathtub curve failure model which applies to hard drives
- SSDs can be characterized as having early an warning phase - which comes
before an early failure weed out phase of the worst drives in the population
and which precedes the onset of predicted endurance based wearout.
this aspect - a small percentage of rogue SSDs account for a disproportionately
high percentage of the total data errors in the population.
report contains plenty of raw data and graphs which can be a valuable resource
for SSD designers and software writers to help them understand how they can
tailor their efforts towards achieving more reliable operation. ...read
the article (pdf)
Let's hope Chris McCall at NexGen is wrong - when he says
PCIe - the Next Infiniband?
Editor:- June 10, 2015 - Don't let
the title of this blog -
the Next Infiniband? - by Chris
McCall, SVP Marketing - NexGen- deter you
from taking a look at it.
PCIe SSD evangelists like
hope that PCIe won't fare like
Infiniband - which -
after a promising birth in 2000 - and
IDC's woefully rash
celebratory plaudit prediction that "more than 75% of all servers shipped
in 2004 would be shipped with InfiniBand connectivity" - became sidelined
by other technologies and lives on mostly as a trusted RDMA veteran in
HPC and consults sometimes as a plush prototyping technology in risk averse
SSD clustering concept proof demonstrators.
No I hope that PCIe as a
fabric will do very much better than IB.
I suppose you could say that
NexGen (the foundling
company discovered in the customer printout and adopted by
Fusion-io but then
later placed back on the pavement outside the car park by
SanDisk - which didn't
want rusty magneto storage sullying its own visions of
purity) has had some skin in the game although more as a systems integrator of
PCIe SSDs than as an incubator.
And for those reasons Chris's blog
provides a clear angst free mini modern survey of the state of the market
ideas about PCIe fabric and lists some useful resources. ...read the
< 10% of FC SAN sites rely on 3rd party benchmarks
June 9, 2015 - Load DynamiX
(a storage performance
testing and validation
company) recently released the
of a survey (pdf) characterized by heavy users of FC SANs (71%) and 2PB or
more of data (76%).
the findings in this set of 115 participants:-
- over half (54%) planned to add all flash arrays to their storage assets in
the next year
- one third (34%) used custom performance scripts as part of their pre
purchase and deployment evaluations
- users were heavily reliant on their current and potential vendors for news
about new products and technologies - and nearly twice as likely (83%) to
rely on news from vendors compared to online magazines (44%)
- 36% said they know and understand the I/O profiles of their currently
deployed tier 1 apps
Having reported on the FC SAN
market for 21 years - I'm not surprised that enterprise users with these
assets have a healthy skepticism about the value of artificial benchmarks when
it comes to evaluating the performance of potential new systems. |
of the findings in Load DynamiX's survey - that only 36% in this highly
technical group say they know and understand their I/O profiles is
The obverse - that most enterprise users
don't know what their I/O workloads are or how they will change after
new purchases - was the catalyst behind the rackmount flash industry's
innovative marketing switch towards risk reducing pricing models (decoupled
from hard and fast analytic performance justifications) which I discussed
the Astrological Age of Enterprise SSD Pricing
Nantero gets $31 million funding for 300º C retention
Editor:- June 2, 2015 -
a $31 million Series E financing round for its
NRAM technology which the company
says is scalable to below 5nm and which has >1,000 years retention at 85º
C or more than 10 years at 300º C.
Nantero was founded 14 years ago, and the last time I wrote about them was in
But the size and educational sophistication of the SSD ecosystem
today means that designers
(and investors) can
appreciate the nuances of difference which might be useful in extreme boundary
Offering a scalability roadmap below the current
commercial limits of flash,
and ruggedness way beyond flash - Nantero's technology has attractive features
which might lure SSD designers out of their 40 year comfort zone of
trapped charges in semiconductor cells.
SAS SSD market shipments beefed up by DWPD-lite models
June 2, 2015 - Commenting on the
SAS SSD market - Don Jeanette,
VP - TrendFocus
says in his new blog -
SSDs continue to show strength in an ever increasing competitive market - "SAS
SSDs are not getting squeezed out by the incursion of SATA on the low end and
PCIe on the high end as many have thought would happen."
explains that SAS SSD makers have populated their product lines with value
models which have much lower DWPD
ratings than the headline performers - which can go some way towards
cost competing with
SATA SSDs while at the same time validating the higher prices of 25 DWPD etc
users don't have to compromise
when they choose SATA
SSDs in their value engineered arrays.
In a conversation with
Marketing Director - EchoStreams
(a whitebox storage company) - in January 2015 - I learned that one of the
design elements in their 20 bay 1U and 48 bay 2U systems - for their customers
who want to build high density flash arrays using COTS SSDs - is that
EchoStreams have deployed
on their drive bays which allow COTS SATA SSDs to be used in the same way as
dual ported SAS but at lower system cost.
Micron in production with 2D 3 bits per cell 16nm nand flash
June 2, 2015 - 2 years after sampling its
16nm nand flash - which was 2D with MLC nodes (2 bits per cell) -
it has progressed to the next evolutionary step and is now shipping 16nm
(which is still 2D) but is now 3 bits per cell (TLC).
In both cases
the products were 16GB memory chips.
Micron says it believes that
TLC will account for almost 50% of the total NAND gigabytes shipped in 2015.
Roadmap for NAND Flash (a classic article) by
Violin's revenue continues to slide
Editor:- June 2,
2015 - Violin Memory
that revenue for the recent quarter ended April 30, 2015 had declined 33% yoy
to $12 million and was 41% less sequentially compared to the previous
Editor's comments:- part of the tragedy of the Violin
story is that so many great things were once expected of it - as you can see by
this news coverage -
4 years ago -
about a $40 million funding round and aspirations to become the "the
fastest growing storage company in the decade".
expectations owed much to its emergence in the market in
2007 when the
rackmount SSD market was at the tail end of the
RAM SSD dominant era
and about to make the fateful transition to the dominance of flash.
Violin did launch its first flash array in
2008 it was a
year behind the rackmount flash SSD market leader in shipments - but that
didn't appear to matter so much in a market where there were so few competing
suppliers. At that time Violin was one of just two leading
flash array suppliers in the early fast rackmount flash SSD accelerator
But the market didn't remain simple. It grew through 2009 (year of SSD market
confusion), evolved into 2010 (year of the enterprise
flash bubble) and then swiftly therafter in many stages became
As all that was happening - Violin's inability to
redeploy its controller technology to lower performance applications and its
unwillingness to focus on a smaller number of winnable segments - led the
company in later years to dissipate its marketing resources in many
unproductive directions - precisely when - due to the growing number of
hundreds of viable competitors - a more focused strategy might have been
Hedvig has $30 million to fix broken SDS market
June 1, 2015 - Hedvig
- which operates in the SDS market - today
an $18 million Series B funding round (bringing the company's total funding to
date upto $30 million).
founder - Avinash
Lakshman - who is credited with building some of the most successful
distributed systems in the world, including Amazon Dynamo, the foundation of the
NoSQL movement, and Apache Cassandra for Facebook said - "We've identified
the potential in a broken and fragmented storage market, and are not only
looking to bring software-defined storage mainstream, but fundamentally change
how companies store and manage data."
I was curious about the origin of the company's name (because I am interested
in such things)
so I asked if the company name "Hedvig" had anything to do with
Harry Potter's owl - Hedwig
(spelled differently but sounds similar).
Instead it seems the name is
an "acronym for Hyperscale Elastic Distributed Virtual Integral Granular
all qualities Hedvig was designed around."
Ah well. Perhaps I
should have guessed.
But I had missed
emergence from stealth in
- what is Hedvig about? And should you care?
As is the case for
nearly every enterprise software company nowadays - SSDs are essential
enablers of what Hedvig does but Hedvig is an SSD-ecosystem company rather
than an SSD-inside company.
Nevertheless - digging around for the
best single document which describes what they do in an SSD context I found
scale-out big data apps need new scale-out storage (pdf) - which among other
things shows where they think their kind of solution fits into the current
sprawl of generationally overlapping and mixed heritage, datacenter
|It's a valuable guide to this
flavor of SDS thinking even if it's not in your immediate plans - because this
kind of philosophy - which we've been seeing coming from many other
sources in recent years - will have a trickle down effect and influence the
design of other types of products too.|
- Less than 9% said that flash already acccounts for 40% or
more of their storage. And nearly half of all participants said that flash is
less than 10% of their storage capacity.
- Over 70% have flash in their budget in 2015.
- 16% of those who had used flash felt they hadn't got the apps
acceleration they expected.
comments:- You can interpret these results
different ways. I see it as showing that there's still a many times bigger
future market for enterprise flash compared to what has already been
- 19% said that storage failures had caused unforeseen outages.
What happened before? - See the
SSD news archive
|Why can't SSD's true
believers agree on a single shared vision?|
|the SSD Heresies |
High Availability Thinking in Pure's Flash Arrays|
|Editor:- June 7, 2015 -
Building Fast, Highly-Available Enterprise Flash Storage from Commodity
Components (pdf) by authors at Pure Storage
describes several interesting aspects of Pure's flash arrays which internally
use consumer grade
The paper presented at
SIGMOD 2015 says among
- Purity can tolerate the loss of 2 SSDs without losing availability. Pure
encourages potential customers to pull drives and unplug controllers as part of
- Due to efficiencies
in deduplication Pure's customers on average provision approximately 12x
more virtual space than physical storage.
- Commenting on the
in capability between flash management which is possible seen from a single
drive level and a global array level the authors say - Pure's controller has a
global view of the workload and much more computational power than the SSD FTL,
allowing it to apply optimizations and make global decisions that the drives are
Pure says that "improvements" in consumer solo drive
benchmarks do not
always follow through to deliver better performance in the managed enterprise
array context. Sometimes the optimized drives perform worse in a Pure array.
- Re quality of service Pure says that to avoid application failures
during controller failure, they have to guarantee that recovery will complete
in under 30 seconds.