- editor mentions on StorageSearch.com and
Who's who in SSD? - LSI
editor - November 18, 2013
For the past year or so I had been wondering
if the glory days of LSI's SSD controller technology lay mostly in the past. But
I can now understand why it took them so long to complete this new design -
which is almost at the integration level of "SSD market in a chip".
commenting recently in SSD news
Editor's earlier comments:- October
LSI - ranked
the top SSD companies in
Q3 2013 - operates in the
SSD controller and
auto tiering / SSD ASAPs
The company sampled its 1st SSD product in
It was a
PCIe SSD - which is
software compatible with SAS
- an interface which LSI
helped to pioneer. Elements of the IP in the new SSD design came from
acquired SandForce in
January 2012 -
and revealed the strategic significance of this shortly afterwards when it
revealed that LSI's PCIe SSD product - the
- was being oemed by EMC.
LSI's SSDs are
than big) SSD architecture and span a wide range of market applications
SSDs and consumer
LSI's interim approach to
ECC IP for SSDs - revealed in August 2013 - is to use tactical LDPC - as a
try it again strategy for improving data usability.
In March 2009 -
LSI announced better
support for flash SSDs
in the latest update to its
SAS adapters. LSI calls this new feature SSD Guard - which can anticipate
some types of flash SSD failures in
RAID 0 configurations
and starts rebuilding data on a spare unit.
2009 - LSI
announced it is
the LSISAS2208 dual-core 6Gb/s
IC to OEM customers. It's
intended to support the forthcoming PCIe 3.0 specification, currently under
development and provide performance levels that meet the needs of
next-generation server platforms based on
flash SSD storage.
The new LSI SAS ROC will deliver performance levels of up to 600,000 IOPS.
they have collaborated on designing
PCIe SSDs for the
enterprise accelerator market - which started sampling in March 2010.
At that time - LSI was approximately the 163rd
company to enter the
SSD market (not counting
SSD SoC makers - which
would push the score to about 185).
in November, 2010 -
published a sponsored
report (pdf) which compares the performance of
HDDs in a simulated
web server environment when managed by LSI's
software - which provides
Editor's commnents:- The report shows that
throughput and access times were improved by at least 3x using a single
SSD cache compared to the HDD only situation.
However - it's
disappointing that the sizing of the test was not best chosen to draw
meaningful conclusions. Because the web content was only 25% larger than
the SSD capacity! It would have been more helpful to design a simulated case in
which there was at least a 10x or 100x size difference. Because if you
can fit all the web content onto an SSD then you don't need the burden of the
"cache" software at all - and might get better results by switching it
There are case studies going back nearly 10 years which show that
SSDs can provide big speedups in web servers. The exact speedup depends on how
fast the SSD is. This test report doesn't answer the question - is LSI's
CacheCade useful in a realistically scaled environment?
In March 2011 -
LSI finally spun off the
business - selling it for $480 million to
LSI was one of
several compatible companies named in
FlashSoft's launch of
its auto tiering SSD
In October 2011 -
a definitive agreement to acquire
approximately $322 million. The transaction is expected to close early in the
first quarter of 2012. SandForce president and CEO, Michael Raam will
become General Manager of LSI's newly formed Flash Components Division.
it has completed the acquisition of
SandForce. And LSI
that its PCIe SSD product - the
- will be oemed by EMC.
2012 - LSI
details of its new
family of SSD technologies - which integrate and join up several previously
standalone elements in its product line in a new unified marketing roadmap.
demonstrated its SandForce
SF-2000 flash controllers working with Toshiba 19nm and Intel 20nm NAND
flash memory at Computex 2012
in Taipei, Taiwan.
In June 2013 -
controllers aimed at the
market. These are:- Opal compliant
SLeeP (pdf) a very low power sleep mode for
|Proximal Data launches
AutoCache for PCIe SSDs|
|Editor:- July 23, 2012 - Proximal Data
immediate availability of its first product - a
SSD ASAP - designed
to work with PCIe SSDs
- and in particular those from
for cache sizes less than 500GB) reduces
in virtualized servers to increase VM density,
performance. The company says it can increase VM density upto 3x with
absolutely no impact on IT operations.
"LSI and Proximal Data have combined their respective solutions
to provide accelerated enterprise storage performance in a virtualized
environment," said LSI's director of worldwide channel sales and marketing
Blanchard. "Proximal Data's AutoCache, when used with the LSI Nytro
WarpDrive PCIe flash card, delivers explosive performance and scalability by
lowering data access latencies and resolving the VM density issue that
challenges virtualized environments. We are excited to bring this combination to
Editor's comments:- here are some
questions I asked about the new product - and the answers I got from Rich Pappas,
Proximal's VP of sales and business development.
Editor:- How long
does it take for the algorithms to reach peak efficiency?
Pappas:- It varies by workload, but typically it takes about 15
minutes for the cache to warm to reach peak efficiency.
the caching only on reads, or is it effective on writes too?
AutoCache will only cache reads, but by virtue of relieving the backend
datastore from read traffic, we have actually seen overall write performance
improvements as well. This effect is also dependent on the workload.
|LSI announces a new
technology roadmap for SSD accelerator components|
|Editor:- April 2, 2012 - LSI today
details of its new
family of SSD technologies - which integrate and join up several previously
standalone elements in its product line in a new unified marketing direction.|
particular LSI is saying that its legacy MegaRAID controllers and software stack
can be used as reliable proven launch pads for its
SSD ASAP /
acceleration software - which is being integrated in new upcoming generations of
PCIe SSD cards (now called Nytro WarpDrives) which use
comments:- in a 2009
storage market forecast I said - "the high end of the
RAID controller market
is going to disappear" - and I explained why companies in that market -
like LSI had to migrate to PCIe SSDs and SSD systems array technology such as
SSD ASAPs to satisfy the emerging needs of their oem customers - which in
previous decades had been met by RAID adapters and controller chips.
LSI has done in the past few years is acquire or develop individual pieces of
the technology puzzle - and selling their storage systems business
Engenio 12 months ago
so that they didn't compete with their storage oem customers - was just as
important as acquiring SandForce.
spoke to LSI about the new Nytro technology last week. From the sales point of
view they see this as offering affordable SSD acceleration for the masses. So
you're going to see low price point fast-enough SSD ASAPs - rather than the
Other common features in the product line are that the
products are bootable, work with legacy
SAS software and
have minimal load on the server CPU.
LSI will also work to get better
integration between the functionality of its SSD controllers and the host cards
and caching software. That should lead to better latency and reliability in the
difference between LSI and FIO?
What's the single
biggest difference you may ask - between LSI and some of the other companies in
this part of the PCIe SSD ASAP market? And in particular a company like
technical ingredients above are very different - and I could summarize that by
saying LSI is at heart an SSD hardware company with most of its IP in chips
- whereas FIO is at heart an SSD software company which uses chips as
deliverables - but nearly all FIO's IP is in software. That's one way of
looking at it - but the clearest difference I see between LSI and FIO is where
they are in the philosophy of their thinking re the SSD market adoption
All the Nytro marketing orientation materials I saw
still talked a lot about how SSDs would fit into an HDD world.
questioned that - I got the impression that LSI's corporate marketing hasn't
gone much beyond that stage. LSI is still at the "SSDs help HDDs point"
whereas FIO and many other SSD makers - and this publication - and many of you
too are beyond that and know that the
future of all
enterprise storage is solid state. The tricky part is navigating safely from
here to there.
Finally - Nytro sounds like a good name for an SSD
brand - but it's not entirely original.
|How big was the
thinking in this SSD's design?|
|Does size really does matter in SSD
By that I mean how big was the mental map? - not how many
inches wide is the SSD.
The novel and the short story both have their
place in literature and the pages look exactly the same. But you know from
experience which works best in different situations and why.
it comes to SSDs - Big versus Small SSD architecture - is something which was
in the designer's mind. Even if they didn't think about it that way at the time.
||For designers, integrators,
end users and investors alike - understanding what follows from these simple
choices predicts a lot of important consequences. ...read the article|
|SSD sudden power
loss vulnerability guide|
|Why should you care
what happens in an SSD when the power goes down? |
This important design
feature - which barely rates a mention in most SSD datasheets and press releases
- has a strong impact on
SSD data integrity
This article will help you understand why some
SSDs which (work perfectly well in one type of application) might fail in
others... even when the changes in the operational environment appear to be
|LSI says it pays to get a
2nd opinion from LDPC |
|Editor:- August 13, 2013 - in a presentation
today at at the Flash Memory
Nibbles and Bits of SSD Data Integrity (pdf) - LSI explained why
reserving the use of LDPC to deal mostly with read error retries (and also
later in the operating life of flash cells) can be a pragmatic design choice.|
instead of applying different strengths of
ECC for fixed
physical block sizes - the company says another approach is to have variable
sized virtual blocks - which effectively means that better cells carry lower ECC
my SSD cell care scheme is
better than yours
Nytro - positioning update|
|Editor:- June 26, 2013 - I had a useful
conversation last week with Rob Callaghan who
manages outbound marketing in LSI's Accelerated
Solutions Division. He wanted to brief me on an
about LSI's family of PCIe SSDs. |
I learned:- one new thing they
do, one thing they still don't do, confirmed something I thought I already knew,
and realized that a headline I had written 3 years ago on this very page - was
wrong. ...read the LSI
interview article in SSD news
| LSI's new
and the petabyte SSD shelf
|Editor:- June 4, 2013 - LSI did a really
good job leveraging their
the impression I got when I was talking last week to Kent Smith, Sr. Director of
Product Marketing who wanted to talk about the
controllers aimed at the
market. These are:- Opal compliant
DevSleep technology (a very low power sleep mode for
SATA SSDs which I
wrote about in an earlier news story lower down this page).
I latched onto in LSI's presentation was that according to
3rd party analyst reports
LSI's SSD controllers were used in approximately 1/3 of all the flash memory
deployed in SSDs in the client and enterprise markets in 2012.
says that even before the start of the current quarter - the SSD world had
consumed over 21 million SF
Now the likely productivity advantages of enterprises using SSD based
notebooks were already known
the SSD notebook market began. And the benefits of having encrypted drives
to reduce the cost of exposure to data loss - when a pc goes astray in an
airport or is stolen - are exactly the same for SSDs as they were for hard
drives - and therefore need no repetition here.
thing for me about LSI's new consumer market controllers - as I said to
Kent - was the new possibilities that they could open up in really high
capacity enterprise arrays used in solid state archives which will eventually
replace disk backup and VTLs.
my roadmap to the
petabyte SSD article (March 2010) I observed that one of the missing IPs in
the SSD market (at that time) which would be needed to implement petabyte scale
physical flash storage in 1U or less - was fast boot SSDs with very low sleep
power consumption. That will enable bulk storage SSD architects to pack an SSD
array into the smallest possible physical volume - and leverage a
tape library type access
architecture at a lower cost
of ownership than tape or hard disk. But to be useful in a solid state world
- the worst case access time would need to be much faster than the 1 to 2
seconds which was the power on ready time for flash SSDs at the time of writing
LSI's implementation of DevSleep already gives a 400x
power reduction in the not needed mode - and Kent told me their power up ready
time is about 250 milli-seconds. In my view that's a good enough figure for
software architects to start planning around - and it doesn't take much of a
stretch to see how that may evolve to get shorter in the next couple of years
- if the market puts a premium on this feature.
The target for the
bulk storage SSD should be to get 5 to 10 petabytes of virtualized flash into
1U of rack height - at a power consumption level which means that every shelf
can have an identical storage density. I had a gut feel it should be do-able -
and an attractive market proposition in the 2016 to 2020 timeframe - based on
the model I published back in 2010. In some ways it should be easier now -
because I didn't anticipate just how good the
technologies for MLC and TLC would get - in particular due to the benefits
So - if you have shares in a company which makes tape libraries or
disk to disk backup - you've already had many years advance warning that those
products will cease to be commercially attractive when the solid state library
market gets going.
|we're #2 in PCIe SSDs and
growing fast - says LSI |
|Editor:- May 15, 2013 - LSI today
it shipped over 40,000 PCIe
SSDs in the past 12 months - and has been ranked the #2 merchant supplier
of enterprise PCIe SSDs in the US, and is the fastest growing vendor in this
category according to a recent report by Forward Insights.|
|new LSI blog on the value
of enterprise flash|
|Editor:- March 14, 2013 - You
won't be surprised to see me mentioning a
published blog by Robert Ober,
System and Processor Architect, LSI - about the
value of PCIe SSDs in
big datacenters - which includes these statements:-
- "Work/$ is the correct metric (and not crazy expensive $/bit)."
I'm guessing that the title of Robert's
blog - What are the driving forces behind going diskless? Will 100% flash
storage make sense in enterprise? - was either inspired by
(stuffing the title with value-loaded words for search-engines) or was
predetermined before the blog was written.
- "when users say - $8k PCIe card in a $4k server really? - I am
always stunned by this"
I prefer this alternative
title - suggested by a banner graphic in the blog itself -
An $8K PCIe
card in an $4k server - huh!?!
LSI do that "my software loves your SSD" thing|
|Editor:- March 6, 2013 - LSI today announced
WarpDrive (PCIe SSDs)
have been validated for use with NetApp's
Accel (SSD ASAP)
Editor's comments:- According to
- "Flash Accel has the ability to keep the cache warm and coherent in
the event of disruptive operations and restart caching from the reboot/crash
point, rather than restarting from a cold cache."
But it's not as
unique in these respects as their document would have you believe - although
this suggestion is probably because of when the document was written.
says in its press release that its "advanced off-loaded
multiprocessor architecture uses up to 4x less CPU and memory resources
than competing solutions".
Now when you see that phrase - off-loaded
- in this kind of context - you can be sure that it's a dig at
The pros and cons - in architectural
aren't as straightforward as they appear from this subliminal value-loaded
phrasing. I discussed these issues a few years ago in an article in
FIO's product page here
on the mouse site.
The motivational reasons you might choose LSI
rather than FIO (or the other way around) probably have less to do with whether
you understand or like the way they design
(which are evidence rather than motivations of what lies behind their
thinking) and instead I think the reasons you might prefer one or other as a
strategic supplier would have rather more to do with whether you're
comfortable with their different
philosophies about the best routes to the
of enterprise storage and, in particular, whether you agree with their
speculation of what the
destination looks like.
If you're going to be in the same waggon
train for 2-3 years - bumping along comfortably together is what's
capacity in SSDs - article by LSI|
|Editor:- January 8, 2013 -
SSD over-provisioning - is the title of a new article published in EDN and written by Kent Smith, Sr. Director of
Product Marketing at the SSD controller part of LSI.|
article describes the trade-offs between performance, the percentage of
over-provisioned flash capacity and the useful impact of compressible data -
which inside SandForce controllers is leveraged to create additional
over-provisioning. The interaction between write amplification counter-measures
and the benefits of using TRIM commands on performance are also noted. ...read
Editor's comments:- there wasn't anything
new for me in this article - which covers similar ground to my 2011 article -
capacity - the iceberg syndrome - which shows how SSD makers leverage
capacity to tweak reliability and performance.
But - having said that -
I learned about over-provisioning by 10 years of talking about it - with
many SSD companies. And some of the things I put in my own article had been
gleaned from past conversations with Kent Smith himself when he was at
SandForce - as well as various other people in
Texas Memory Systems
guessing that what Kent would have liked to say on OP may have been "trimmed"
by a word count limit in his latest EDN article.
So here are some
other suggestions for more substantial and ideas packed articles I recommend
- which Kent Smith has written in the past for other publications, and which
cover SSD controllers from other angles:-
|LSI ships 1
million SandForce controllers / month|
|Editor:- July 31, 2012 - LSI has announced
enhanced support for the
market in its SandForce
SF-2200/2100 controllers:- enabling lower SSD power consumption, faster
boot and support for "virtually all MLC flash product families".|
has shipped well over 10 million
and we anticipate our shipment volumes will continue to increase, driven by the
exploding demand and lowering price points for NAND flash technology," said
VP of marketing, Flash Components Division, LSI.
comments:- last week I asked LSI if the improved power saving feature was
related in any way to
care. I haven't got an answer yet - so it may be the answer is No.
On the other hand maybe they're waiting for the
Flash Memory Summit (in 3 weeks
time) before they say more about their adaptive write DSP IP roadmap.
||LSI/SandForce have shipped over 10 million SSD
controllers - since
2010 - and
they're currently shipping over 1 million per month.|