|
|
| .. |
|
Editor:- June 2016 - FlashSoft was one of those rare SSD
companies whose technology and brand continued to be developed many years
after having been acquired. (This is being written 4 years after FlashSoft was
acquired in February
2012.)
And I believe the insights and contacts in the enterprise
SSD market - which came from having direct contacts with users of its enterprise
software at a critical period of change helped its purchaser -
SanDisk - to evaluate
important user trends in the market and better adapt towards an enterprise
friendlier way of thinking.
How much of that - if any - will feed
through into adoption by SanDisk's new owner -
Western Digital -
remains to be seen.
From a legacy point of view it would be much
simpler for WDC to seek hybrid
and cloud-like systems
level software by another acquisition - such as
Tegile - rather than
integrate all the various software jigsaw puzzle parts which come with the
component level IP sets of SanDisk (which include Fusion-io and Schooner
Information Technology for example).
A critical test is how WDC plans
to position itself within the emerging
SSD and SCM
DIMM wars market.
This is market it could have chosen to stand
aside from - before it acquired SanDisk. But the big memory market is central to
the future profitability of flash (and WDC now has skin in that game) and the
big memory market also intersects and competes with product lines (such as
PCIe SSDs and 1st
generation memory
channel SSDs) in which SanDisk has had iconic products in the past. (I
expect some of those legacy product lines to be end of lifed.)
SSD aware software -
rather than interface and controller chip design - has become the new heat pump
in the entropy of data systems. The outline challenges and complexities of the
big memory market are still not widely understood. That's why I made this
difficult theme the subject of my recent home page blog -
where are we heading with memory
intensive systems and software? |
| . |
 |
. |
|
|
by
Zsolt Kerekes,
editor, StorageSearch.com
- December 2011
FlashSoft is 1 of 30 companies
which has launched
SSD ASAPs, and
was 1 of the first 2 software
companies to ever be listed in StorageSearch.com's long running
Top SSD Companies List
- based on reader metrics in Q3 2011.
As you can see from the company's
current description of itself (above) it's clear today what the company is
doing / aspiring towards.
But when Ted Sanford, CEO of
FlashSoft first contacted me (while in stealth mode and more than a year
before initial
product launch) from his cryptic gmail address - to discuss the
market for his new auto tiering / SSD ASAP software company - my immediate
(mistaken) reaction on seeing his email was - "Oh is this going to be
another company looking at
hybrid
notebook SSDs?"
But
I perked up when he went on to explain that - no it wasn't about notebooks
- their product was aimed squarely at the server acceleration market.
From
what I learned then - it seemed to me that they might be able to make a
significant difference to the dynamics of how SSDs were being sold and used. If
their product worked and if their marketing was worked too.
I knew
that lots of other companies were working on this problem too and competition
would be tough. Over 30 companies have since
announced SSD ASAP
related products - and I think this type of technology will eventually be
marketed by hundreds of vendors.
In the 2 years since the earliest SSD
ASAP products came to market - I've
upgraded my view
about where they fit into the storage landscape.
I now think these
products will have a permanent place in the storage market –
instead of merely being a stop-gap product (while HDD arrays are still in use in
the enterprise).
FlashSoft was the first company I spoke to which had
also independently reached the same conclusion.
It's unusual to
suggest that a startup might have a vision of where its technology would fit
into the SSD market in 5 years' time - but from the strategic point of view - I
think that unless SSD companies are clear about that kind of thinking right from
the start - they will often be reactive and off target and waste resources
in directions which prove to be futile.
I used the term "its
technology" to differentiate between technology and the legal entity which
is the company. The stats from
storage history
indicate that successful new storage software startups get acquired - while
most of the rest go bust.
How
should you view SSD ASAPs?
The mental model I'm offering
StorageSearch.com readers today is you should think about ASAP IP in the same
way you think about SSD
controllers. There are useful analogies to be drawn from that market.
Although
there will be a small proportion of SSD oems who develop or acquire their own
proprietary ASAP technology - I think most of the hundreds of vendors who
want to be in the SSD market (because it sells more SSDs) won't find it
economic or sustainable to develop their own technologies in-house. Instead
they will source solutions from a merchant market for ASAP software, IP
and chips – for the same reasons that oems buy 3rd party SSD
controllers. (Speed to market, lower risk, lower investment cost, and better ROI
while volumes are still low.)
Within the SSD ASAP market I think
FlashSoft fits into these classifications:-
- small
architecture rather than big architecture.
I'd better clarify what
I mean by "big architectur" in the ASAP context.
In the
case of an SSD controller - the dividing line is the number of flash chips
around which the design is optimized. This affects the entry level product
size and also the scaled up economics (when reliability is factored in) and
theoretical limits to performance (when using the same number of chips in a
customer's rack).
Here in the context of ASAPs I think the dividing
line for big architecture - is one that starts with multiple levels of SSD
storage, connected to tens or hundreds of servers - and
petabytes of SSD
storage. There are no real examples of this product classification which exist
yet. But there will be. The current version of FlashSoft's product
supports a terabyte of flash cache using a
skinny RAM cache
of only 150MB in the server. It uses 3% to 5% of the host CPU cycles. Ted
Sanford told me their next Linux release will support upto 256TB flash.
FlashSoft
has got the usual stuff on their website such as
benchmarks, tech overviews etc.
When I asked yesterday - is there anything in particular you want
me and my readers to know? - the 1st thing on Ted's list was - to please
mention their
October
2011 - release of version 2 software for PCIe SSDs.
He said - "I
know you've heard this info, so I won't repeat it, but we think it's a big milestone
for the company, especially delivering a 2.0 version before any other software
company has brought out their 1.0 version."
Last week - at the
Server Design Summit - Ted gave a
presentation - and he told me - "I think it reflects our vision for
ASAP soltions.
"Today SSD vendors are generally thinking in terms
of a single server architecture. We took the lead in developing caching that
would run in a single server, but as you can see from this presentation, we
built our first products as the "base" of a "technology pyramid"
that would lead to clustered environments, integration between server-tier flash
and storage arrays and virtualization evironments."
He sent me his
presentation along with his speaker notes and I think it gives the
clearest picture of what FlashSoft is aiming at.
You can judge for
yourself by looking at this overview -
Why your next
server will have a solid-state cache (pdf)
For related info take a
look at the links above and -
FlashSoft
- editor mentions on StorageSearch.com.
I currently talk to more
than 300 makers of SSDs and another 100 or so companies which are
closely enmeshed around the SSD ecosphere - which are all profiled here on
the mouse site.
I learn about new SSD companies every day, including
many in stealth mode. If you're interested in the growing
big picture of
the SSD market canvass - StorageSearch will help you along the way. Many
SSD company CEOs read our site too - and say they value our thought leading SSD
content - even when we say something that's not always comfortable to hear. I
hope you'll find it it useful too. |
| . |
|
|
SanDisk (enterprise) was a
prominent advertiser here on StorageSearch.com in the years upto its
acquisition by WDC. The product line they promoted here was the FlashSoft
software IP. You can see one of several ads from the 2014/2015 campaigns here
below. In later years the whole SSD market became much more receptive to the
idea of big memory and replacing DRAM with flash and other alt nvms. See:-
where are we
heading with memory intensive systems? for more about that strategic trend. |
| . |
 |
| . |
|
|
In June 2011 -
FlashSoft
announced it has
secured
$3 million Series A funding and has launched its first product -
software which enables enterprise flash to be used as a cost-effective,
server-tier computing resource (ASAP functionality in
software) which is available for free evaluation through a 30-day "Try
Before You Buy" program.
In October 2011 -
FlashSoft became the
first ISV to enter the top
20 SSD companies list.
In February
2012 -
SanDisk
announced
it has acquired FlashSoft
- one of the leading independent software vendors in the
SSD ASAPs market.
- editor's comments:- October 2011 -
FlashSoft was one of
the first two software
companies to ever be listed in StorageSearch.com's long running
Top SSD Companies List.
FlashSoft was ranked #14 - based on SSD search metrics in the
3rd quarter of 2011. That was slightly ahead of
IO Turbine - the only
other software company to have achieved this accolade. IO Turbine would have
entered the top 20 at #16 in its own right - if it had not been
acquired before the list was published - by
Fusion-io.
FlashSoft
designs a software product which provide
auto-tiering / SSD ASAP
functionality when used with almost any industry standard vanilla flash SSDs.
FlashSoft's tagline for this - which I like - is - "Software that
unleashes the power of enterprise flash". Another way they describe it
is as a "tier minus one" software solution. I'm not so keen on that
description - but I suppose it will stick. See
why I tire of "Tier
Zero Storage"
FlashSoft was founded by a team with decades
of experience at companies including VERITAS, Oracle and Symantec. Having
good connections - FlashSoft has been quick to get off the ground and
establish technical liaisons with a bunch of leading SSD and server
infrastructure partners.
The exciting prospect for vanilla SSD oems
is that FlashSoft's new product - if it works effectively - will make it much
easier for them to sell SSDs to new customers who haven't been economically
attractive to SSD vendors in the past for the reasons I described in my article
-
analyzing the emerging
market for SSD ASAPs.
If it's such a good idea then why
aren't other companies doing the same thing?
Well Microsoft did
try. They decided not to pursue it - and sold their technology to
Adaptec who used it as
the core of their MaxIQ SSD Cache.
Other companies in the software
only ASAP space - include IO
Turbine, NVELO and
NEVEX Virtual Technologies.
I
spoke to FlashSoft's CEO Ted
Sanford (June 29, 2011) about their business plans and technology.
FlashSoft's
value proposition is the classic Auto-tuning SSD Accelerated Pools of storage
(ASAPs) case:- which aims to provides an economic and simple to install tool
to enable servers to get the benefits of SSD acceleration without the costs of
either:- migrating all the data to SSD or the high cost of human engineered
tuning (which can go out of date). The company says it has achieved worthwhile
4x to 5x speedups in SQL apps which are documented on its web
site.
FlashSoft doesn't yet ship software for clustered servers - but
that's in the pipeline.
How much does it cost?
At
launch time a single server license for a fast
SAS SSD is about $1,500
- while a PCIe SSD
license costs a shade over $2k. (Have they tested their product with a
Fusion-io card I hear
you ask? - Yes - among many others.)
I think the product will be
attractive for SSD oems - who can easily tweak the code to make it work better
with their own products and open new markets which would remain closed without
the simplicity of the ASAP concept.
Summary - I started talking
to the company a year ago when they were still in stealth mode. The direction
they've taken and the execution of their launch plan demonstrates a lot of
business maturity - which is rare in software start ups. Many companies in the
past have tried and failed to convincingly break though the technical and market
acceptance barriers in the auto tiering SSD market. This product won't fix all
problems - but if their execution remains on track - I'd say that FlashSoft
has a good chance of doing in SSD software - what
SandForce did for
SSD controllers. It
will make it easier
and cheaper for a new wave of server oems and storage integrators to
expand the reach of SSDs in the enterprise.
FlashSoft will also
work with RAM SSDs
Most of you won't be interested in this bit -
which is why I left it till last. But regular readers may know from my
SSD branding series
that I'm interested in the subject of names and brands in the SSD market - not
just in the technologies and markets. And another thing I said in my meeting
with FlashSoft that I thought they had chosen a good name for the company -
because it describes what they do.
But I also pointed out that
some people may think from the name that their solution is only applicable to
flash SSDs. (Just as the
the wrong signal
gets sent when readers see "RamSan " etc.)
FlashSoft
agreed that their software is not intrinsically restricted to working with
flash SSDs - because they already work with latencies in the 10 to 20
microseconds range with flash (even though with flash -
the write times
are to cache
rather than directly through to flash media). Bottom line there's no technical
reason why their hot spot tuning software shouldn't work equally well (or
better) with RAM SSDs
too.
Once the company has got further along with clustering support
- and supporting more operating systems this could make it easier for RAM SSD
vendors to get into some applications too. Having said that - the economics
depends on the topology of the SAN. If you've got 20+ servers sharing data on
the SAN it may still be
cheaper to put a single specialist SAN SSD appliance in the path to that data
- rather than buy 20 sets of local SSDs and 20 sets of licenses.
In the
long term I think that some customers will do both. Use (less) locally
accelerated servers - and still need faster RAM SSD layers above that. And
even when all storage becomes solid state - something which FlashSoft and I
agree about is that there will still be a vastly different speeds of installed
solid state storage in the enterprise. So the need to manage performance between
the apps level server and the outer layers of bulk storage will still be there
in 5 to 10 years time. A nice kind of problem to worry about - if you're a
start up. |
|
| . |
| Other articles on StorageSearch.com -
which you may be interested in
|
| . |
|
| |
.... |
| after AFAs
- what's next? |
Throughout the
history of
the data storage market we've always expected the capacity of enterprise user
memory systems to be much smaller than the capacity of all the other attached
storage in the same data processing environment.
 A
thought provoking blog on StorageSearch.com
- cloud
adapted memory systems - asks (among other things) if this will always be
true. ...read
the article | | |
| .... |
 |
| .... |
| One cache to bind them |
Editor:- April 2, 2015 - SanDisk - which in the
past 4 years has
acquired 4
companies listed in the Top
SSD Companies List (published by StorageSearch.com)
- recently
announced
the availability of acceleration bundles which integrate 3 of the most recent.
Among other things these include:-
"We
have built our enterprise flash business to offer the broadest range of storage
acceleration solutions to address customers' changing needs," said
Ravi Swaminathan,
VP and GM, Systems and Software Solutions, SanDisk. "The new
hardware-software bundles offer the best of both our hardware and software
portfolios in a single solution."
Editor's comments:- The
new FlashSoft bundles will appeal to some types of enterprise users who like
this style of roadmap continuity - where the personality of their SSD (software
and controllers) come from the same source. | | |
| .. |
|
|
| .. |
| "Many enterprise
users - who wouldn't dream of approaching SanDisk to use its raw SSDs - seem
more than willing to use SanDisk's enterprise SSD software (FlashSoft). Can
SanDisk leverage this to transform itself into an enterprise SSD heavyweight?" |
| Editor's comments:-
SSD news
- June 19, 2013 | | |
| .. |
 |
| Where are we now
with that SSD software? |
| .. |
| is SanDisk really
nurturing true enterprise SSD DNA? |
Editor:- August 15, 2012 - Do you remember FlashSoft?
Many of you still do. It was one of the
top enterprise
SSD software companies
before it got acquired 6 months ago by SanDisk.
One
of the tips in the
Survivor's
Guide to Enterprise SSDs - is that when it comes to SSDs - rules are
made to be broken.
And earlier this week I learned this can
apply to my own gut feel rules of thumb too. The unwritten rule being that
semiconductor companies generally make a mess of enterprise software and are not
so hot at understanding the enterprise SSD market either.
Frankly
I had expected that FlashSoft would disappear into SanDisk - and would get
smothered by a marketing organization which had many times before demonstrated
its lack of awareness of the fundamentals of good enterprise SSD marketing. And
that was the tone of my parting message to the founders along with a few words
of congratulations as they disappeared into the new SNDK afterlife. I never
expected to hear from them again.
So the first thing I asked Rich Petersen -
(former VP of Marketing at FlashSoft and now Director, Marketing Management at
SanDisk) a few days ago was - how are they doing as part of a chip company?
What are they doing with the FlashSoft brand? How do they plan to develop the
enterprise SSD business? etc.
One of the things that Rich had wanted
to talk about was the release of new support in their caching software for
VMware vSphere.
We spent a lot of time talking about that too - and had a big discussion about
the role of SSD software - not only as a business tool - but in effect as a new
way of virtualizing and looking at enterprise SSDs and how they can fit into
architecture models. (My view is that a powerful SSD suite - if it becomes
widely used - can be as significant to the SSD market - as a new interface
or form factor.)
We covered enough ground to write several long
articles. I'm not going to do that today - because I'm supposed to be on
vacation and sitting out in the garden by my pool.
So you should
regard this as the really really short version - and a placeholder for much
more detail which I will return to later.
FlashSoft - or the
enterprise SSD software part of SanDisk (or whatever else you may want to call
it) is today operating in a business mode which is like what you would expect
from a best of breed enterprise SSD systems company. They talk to end users like
they've always done. They learn to change important aspects of how the products
work and are sold because of feedback from end users - and not because they've
read that something is a good idea in a
market analyst's
report.
There are some surprising consequences of this at the
technical and business level.
Chief among those surprises for me is
that FlashSoft says it will still support other brands of SSDs. Rich
explained this was just a pragmatic business decision. Big users told them they
like FlashSoft - but they already use or might want to use non-SanDisk SSDs.
These users are only going to standardize on one SSD software platform. They
don't want to learn 2 different ways of doing the same thing.
On the
other hand an advantage of having access to an enterprise SSD maker is that if a
big user needs some expensive hardware on which to evaluate the benefits of
their software - then it's easier on the marketing budget to get some SanDisk
SSDs to do this.
FlashSoft's visibility into what enterprise end users
really do - and the suprising preferences they have - which are driven by
customer business optimizations rather than simplistic technical extrapolations
- also means that - like rackmount SSD companies - FlashSoft learns valuable
market lessons which can be reapplied to optimize designs in future SanDisk
enterprise silicon. | | |
| .. |
 |
| .. |
| SanDisk acquires FlashSoft
- but... |
Editor:- February 15, 2012 - SanDisk today
announced
it has acquired FlashSoft
- one of the leading independent
software vendors in the
SSD ASAPs market.
SanDisk says it intends to sell FlashSoft's products as standalone software, as
well as offer these software products in combination with SanDisk's growing
portfolio of SAS, PCIe and SATA enterprise solutions.
"The acquisition of FlashSoft represents an important step in
SanDisk's strategy of delivering complete SSD and software solutions to
enterprise storage customers," said Sanjay Mehrotra, president and
CEO of SanDisk. "FlashSoft's software products complement our growing
family of SAS, PCIe and SATA Enterprise SSDs, and we are pleased to welcome
FlashSoft to the SanDisk family."
Editor's comments:- I'm not surprised that someone has
acquired FlashSoft - because they were an obvious target sitting so high in the
Top SSD companies list.
This
means that SanDisk now joins an impressive roster of enterprise SSD makers who
have acquired auto acceleration / virtualization software companies in the
past year.
The
only stumbling block is that acquiring enterprise SSD assets isn't the same as
being able to do anything useful with them afterwards from the business point
of view. Especially when they're software companies.
I still remain
unconvinced that SanDisk has achieved as much as it should have done from its
earlier acquisition of enterprise SSD controller maker
Pliant.
(Although Pliant made SSDs - they had virtually no market share - so their main
value to SDK was as a sounrce of enterprise controller IP.) |
|
|
I don't think SanDisk
understands the enterprise SSD market in the same way as the other companies
with which it competes. It's not the same as marketing consumer products. | | | |
| .... |
| FlashSoft
unleashes the power of enterprise flash |
Editor:- June 28, 2011 -
FlashSoft
today announced it has
secured
$3 million Series A funding and has launched its first product -
software which enables enterprise flash to be used as a cost-effective,
server-tier computing resource (ASAP functionality in
software) which is available for free evaluation through a 30-day "Try
Before You Buy" program.
FlashSoft says that despite the
performance advantages of flash SSD, 2 barriers have inhibited its adoption in
the enterprise.
- First, when used as primary data storage, flash memory cannot easily
integrate with and leverage the benefits of existing storage systems
infrastructure.
- Secondly, storing all of an application's data on server-attached flash
memory remains expensive.
FlashSoft's new all-software product
overcomes both of these objections with what they call a "tier minus one"
solution for flash virtualization. Enterprise IT can now provide databases,
applications and virtual machine environments with the performance benefit of
having the entire data set on flash, with only a fraction of the data actually
stored in flash.
This innovation makes enterprise flash a
cost-effective performance solution that works seamlessly with existing storage
infrastructure. In fact, FlashSoft actually reduces the IO burden on storage,
producing even greater cost savings. FlashSoft's technology is designed to
deliver flash-grade performance within a standalone server, across server
clusters, and throughout the data center. Early Customer Successes
One early user -
Zenprise
said "By using FlashSoft, we aren't buying new server hardware or
licensing additional server software. We're simply making our existing servers
and software run at their full potential." And they were equally equally
impressed by FlashSoft's reliability when they set up stress tests (read case study).
In conjunction with its funding announcement, FlashSoft announced that it is
collaborating with industry leaders including VMware, Microsoft,
SanDisk,
Virident Systems,
LSI,
OCZ, and server maker
AMAX. These relationships will help
FlashSoft integrate its software more closely with complementary hardware and
software products, and provide customers with the best solutions for their
specific requirements.
Editor's comments:- FlashSoft says its
software (which runs on Windows Server - Linux is in beta) works with any
flash SSDs upto 1TB, and takes approx 5% CPU utilization and 100MB of core RAM.
I asked Ted Sanford,
founder/CEO some questions.
How many physical SSDs does the
software support?
The number of SSDs is not limited, as long as they can be represented
as a single logical volume, eg. through a RAID.
Is the 1TB limit shown on your site the limit for the setof SSDs or just for
each drive?
The 1TB limit is the current logical limit for the
SSD used for caching. The data set is typically 5x greater (or more) than the
cache. The size restriction is an artifact of early development, and in a
near-future release, there will be no restriction on the size of the SSD
employed.
In the case of
sudden power
loss what are the steps taken to protect the state of the cached data and
update the external storage?
FlashSoft employs a method called multi-level metadata
management, which stores some cache metadata in RAM, but most of it on the SSD
itself (and employs a balanced tree design for optimal efficiency). There are
two benefits to this design: first, it minimizes utilization of server memory.
Only the hottest metadata runs in server memory. The rest is cached in SSD.
Also, the application regularly creates snapshots of the metadata on the SSD, so
that in the event of a server crash, the cache metadata can be re-created from
the snapshots + most recent metadata almost immediately. |
 |
Typical recovery is less than a second. (Keep
in mind, our team's background is at Veritas, Oracle, Symantec, etc. so data
recovery is a top priority for the product | | | |
| . | |
