 what
changed in SSD year 2013?
the big SSD ideas to assimilate
and the big SSD ideas to forget
lessons from 2013by
Zsolt Kerekes,
editor - StorageSearch.com
- December 9, 2013
Are there any new big changes in the ideas
related to the SSD market and its related technology and business directions
which have emerged in 2013 - and which it would be good for you to think about
when formulating your own plans for 2014 - and guiding your day to day
assessment of products and companies to follow?
Yes.
And in
this blog I'll tell you what I think they are.
I was talking about the
desirability of an annual refresh of SSD ideas in a
recent conversation
with
Rado Danilak
founder and CEO of Skyera.
I
said - "I've been in the enterprise market for a long time and my
experience goes something like this.... Every year I learn 2 new important
new ideas about SSDs. But every year I also have to remember to forget or
discard 1 old idea which was vital to know before because it's no
longer useful, valid or true."
I hope that - whatever your level
of SSD knowledge - you too Dear Reader can either agree with the principle
of this statement - or derive some comfort from it.
In the case of
Rado - who is one of the world's most experienced experts when it comes to
optimizing flash SSD
controller architectures - his list of what's old - and needs to be
forgotten - and what's new and needs to be assimilated - would be different to
yours and mine.
Indeed everyone comes at this from a different
starting point.
So when people starting new jobs or projects
within the SSD industry - approach me somewhat hesitantly to say hi! - and
say - I'm new to SSD - I say - don't worry too much about that - it means
you've got less
old
baggage dragging you in the wrong direction.
What are these
past directions?
If you've seen my
SSD history
article (which is thinly-disguised recycled old
SSD news) you may have
noticed that in the past 6-7 years or so at round about this time of year I've
tried to anticipate and then later retrospectively summarize the themes for
each year.
If you trawl back far enough in time - the number of really new big
ideas and changes in each year was initially just one truly outstanding theme
per year. But as we edge closer to the present day - then because the SSD
market itself has grown (in both technical complexity and the scope of its
market reach) the number of significant twists in this annual SSD narrative
thread has expanded to 2 or 3 and sometimes even 4. It doesn't sound like a
lot. But it's a lot of ideas to digest at a single yearly sitting. (Which is
an argument for dipping into these pages at shorter intervals and snacking your
SSD shock nibblets.)
When I started thinking - what's going to be in
my new SSD idea / emerging trend list for 2013/14? I was coming at it from the
perspective of "new" - as in - not just something which I've already
written about in one of these annual summaries before.
 That
means my reference point for this article - starts with the sum of all the
things I've said before - and will focus on the genuinely new things.
So - in a way - you need to know some of what I've said before -
otherwise you'll be wondering - why did the
SSDmouse not mention
at all something which everyone else seems to be writing about now? (Maybe
because I already told you about those exact same things in earlier years - and
these "new" SSD issues are being rediscovered by new SSD analysts
in the same sense that Europeans "discovered" the rest of the
world.)
But it would be unfair for me to expect you to read all that
previous SSD stuff - especially if you haven't seen it before - because (as I
already warned you at the start of this article) half of the ideas which were
really helpful and important 3-5 years ago - don't help you very much any more.
So
instead - I'm going to take as my starting point a quick recap of what I wrote
about as the main new trends in 2012 - see if they are still valid or not - and
then do my update for 2013.
The advance clue here is that one of the
safe assumptions about the SSD market - which held true all the way through 2012
- is now something which is no longer true - and which you need to forget.
OK
let's recall some of the key changes in 2012 which I highlighted in my
2012 SSD market
transitions article (which originally appeared as a home page blog like this
one.) The important ideas emerging at that time were:-
- diversity in PCIe
SSDs - in particular the new
2.5" form factor.
In 2013 the trend to more ways of using PCIe SSDs expanded in 2 directions.
- adaptive
R/W - which a few years ago was a niche technology - but which has been
finding its way into more products from more companies. For example - in news
stories in the 2nd half of 2013 - this technology was a key factor in 2
acquisitions (WD
buying Stec,
SanDisk buying
SMART), and this
technology was the reason for a technology licensing deal between
Toshiba and
Densbits.
But
you know how technology trends go in SSD...
In 2014 - some version of
adaptive R/W DSP technology will be included within the product lines of nearly
all SSD makers - because - apart from organic development and licensing - it's
one of the design techniques which has enabled
LSI to design a single
SSD controller chip - the
SF3700
- which I described as "SSD market in a chip" - when it was launched
in November 2013 - and which can be configured by firmware and a suitable
supporting cast of flash
memory chips to deliver a very wide range of SSD power consumption and
performance metrics all within a
single
design (pdf).
A lot of SSD makers already use
SandForce controllers.
When they use the new 3rd generation chips - then among other things - they also
get adaptive R/W DSP inside.
- efficiency within
SSD design - emerging as a concept for users (because SSD utilization
affects running costs) and for vendors and
investors (because it
dictates the elasticity of profit margins - compared to competing SSDs with
different design efficiencies).
- more acquisitions
- I promised you a lot more of those in 2013. You got them and you're still
seeing more of them for the reasons described in my article -
hostage to the
fortunes of SSD.
Some people seeing this trend may ask - will this
activity (bigger companies buying smaller companies) lead to a
reduction
of competitors in the SSD market? My answer to that is no - not for the next
2-3 years at least.
My reasoning for this is that no single technology
or set of SSD IP owned by any particular vendor - can optimally satisfy more
than a small percentage of SSD market uses - while also remaining the most
competitive in each of those
segments.
And so for long as that remains true - any big company would have to
acquire maybe 5 to 10 SSD companies to get the technology it needs at any
point in time.
But even if it were to do that - the question of
whether it can understand all these sub-markets within SSD and perform
adequately from a marketing perspective to remain relevant to customers
(compared to each company which it acquired) is very doubtful. That's aside from
the question of how long the hot talent will want to remain in the stifling
cocoon of a bigger organization.
And you can add to that - the
continuing disruptive effect of ever more new SSD companies - and the
growing revenue
value of the SSD market - which will attract more competition for as long
as anyone thinks they can do a better job than all the other companies already
in the market.
- the potential of SSD
software to become the most important gateway for managing everything.
Buried
among the many similar sounding news stories which paired the words SSD and
software in 2013 - there have been some gems of difference visible too.
- ratios
of things within the SSD market - is becoming a useful business
benchmarking indicator for performance and economics. You don't have to be an
SSD-head to understand that the relative ratios of parameters in arcane SSD
technologies can tell you something useful. Within this ratios idea - I
include many different ideas including:-
- how many directly attached SSDs should each server have? (for different
markets and different apps setups)
- ratio of SSD capacity -
how much server
attached SSD to that on the SAN? (for legacy architectures)
- ratio of SSD capacity - between different speeds and hierarchies of SSD
rackmount (ultrafast,
fast, fast-enough, cloud etc)
This also includes the ratios of
apps server racks compared to storage racks - because in my enterprise model the
apps server - is simply the base level of this hierarchy. However, in some
systems architectures - it may be the only level.
- ideal ratios of new SSD
cache to legacy HDD array capacities in legacy and transitional systems.
- how many
flash chips to a single controller? - (in different markets)
- how many chips in an SSD controller? - (for different types of SSD array)
- how many SSD vendors should any single SSD software platform support? - Is
more - always better?
- ratio of RAM
to flash in SSD caches and exteranlly in systems
Anyway
- all the stuff above - had already been discussed before 2013.
what's
new as we look back on 2013?
Here's my shortlist of new SSD ideas
to incorporate into your thinking based on what's been happening in 2013.
An
important new SSD idea to remember - re enterprise SSDs - is the
rack has
become the most important form factor at which level enterprise SSD vendors
must focus their strategic product ideas.
What we're seeing in the
market today at the rack level - are efficiencies and competitive advantages
which accrue from combining and integrating design factors at many levels
within large SSD arrays (at the memory utilization level, the SSD controller
level, the drive interface level, the flash array organization level and
multiple levels up and down the system software and apps software stacks).
Mastering the design possibilities of SSD at the rack level enables new levels
of competitive advantages for vendors.
Moreover, participication in
the market at the rack level - can enable vendors to learn how users value the
features they find in SSD racks in unexpected ways. By leveraging this usage
information for specific use cases and market segments - vendors of rackmount
SSDs can - if they choose to do so - make their products uniquely attractive at
price points which cannot be achieved by standard arrays of vanilla SSDs.
The
importance of lessons learned at the rackmount SSD level - as a business
development strategy for SSD drive vendors cannot be overstated in this phase
of the market. Although many of these lessons can be learned indirectly by
other means- the risk of being remote from the user experience of deploying
large SSD arrays - places drive vendors at a disadvantage where what they do
isn't seen to be unique - and their prime route to market is as a commodity.
 A
new SSD idea to forget - is traction in the Top 10 part of the
Top SSD Companies List
(StorageSearch.com's list of leading companies based on search volume).
In
the past 2 years it was rare for companies to break into the top 10, and rarer
still for newcomers to break into the top 5 part of this list. So a safe
working assumption was that companies who had achieved those dizzy heights of
technology, brand and thought leadership were consistently doing things better
than all other SSD companies - and so were a prime choice to include in long
range strategic plans.
The change I predict - is that in 2014 we'll see
more new names at this end of the list - for 2 reasons.
- Companies in this list have always scored strongly as acquisition targets.
In the past 8 quarters - out of the 19 companies which have had at least one
appearance in the top 10 section of the list - 7 have been acquired.
Of
the top 20 SSD companies named in the most recent quarter alone - 5 have
already acquired in the 2 months since that list was published.
As
we've seen in the history of this list - the companies doing the acquiring
aren't guaranteed to immediately inherit this attribute from the companies they
buy. This creates a vacuum of uncertainty which facilitates shuffling.
- People at the leading edge of the SSD market today have a much stronger
conceptual grasp of what the SSD market can do for them - than ever before -
simply because SSDs are a much bigger part of their every day work and thinking
experience.
Consequently new product ideas which solve genuine
problems in a useful way - can be understood, assimilated and have a market
impact sooner today than ever before.
And another thing which will
accelerate the sudden appearance of completely unknown companies into easy name
recognition is the fact that - unlike previous times in the SSD market - where
new companies had to develop almost everything for themselves - today new
companies entering the market can rely on a sophisticated SSD ecosystem
- in which key elements of their solutions are already being supplied by other
companies. So all they need to do to get to the next level is to leverage that
ecosystem - and also leverage the fact that the problems they solve are already
becoming widely known. New form factors in SSD
- in the enterprise
- memory channel
SSDs - by which I mean flash SSDs which can provide strategic apps
enhancement functionality and which plug into in legacy DRAM slots - have
become a serious
business proposition.
The most significant product in this form
factor has been the architecture designed by
Diablo Technologies -
which was brought to market by
SMART Storage
(which is now part of the enterprise SSD business at
SanDisk).
It's
likely that other memory vendors will also step up their SSD efforts in this
form factor if enough server makers adopt the concept.
- in the consumer
market - users have now got new reasons to get excited by SSDs in notebooks
- as they can get either ultra low power (with a new SATA sleep mode) or faster
performance (with PCIe SSD speeds) within the same single form factor (M.2). LSI
has a reference design SSD which can satisfy both markets with the same card
design and same controller. A single link is all that's needed for oems to
determine which personality this SSD will have when it gets shipped. The
availability of low latency PCIe SSDs in the consumer market in 2014 may
A
new emerging enterprise SSD idea for 2014 will be the idea of "SSD
software platforms" which will promise to solve a wide range of SSD
integration and utilization problems from the server SSD (single drive) to
the SAN SSD (rack) level with a single software architecture which
supports multiple SSD types over many product lifetimes.
These SSD
platforms can be viewed as a new strategic lens through which enterprise
users can think about and plan future SSD deployments independent of the
SSD itself.
In the past 4 years we've seen SSD-centric software change
from being almost non existent, then growing in importance as a sales
acceleration tool to simplify the deployment of more SSDs. You could almost
say that some types of SSD software functionality - such as
caching - have
become almost commodity-like in what they appear to offer.
My gut feel
is that - just as there's no single type of integrated SSD box which is ideal
for all SSD applications - there probably isn't a single "SSD
software platform" which will ideally suit all use cases either.
Nevertheless
- it may be that some platforms could dominate some types of applications.
Eventually
- in 5 years or so - it will be obvious who the 4 or 5 SSD software platform
heavyweights are. And it's possible to make some guesses about that now. But
some of the companies in this space have only recently emerged from stealth -
while others haven't even announced their first products yet.
In
2014
we will see the battle lines for the SSD platform being drawn up - as ISVs and
SSD drive makers and SSD systems companies all try to convince you that any long
term plans you make will be more future-proof if you use their software.
Regardless
of which handful of companies wins the ultimate battle of the SSD platform in
2018 - inevitably that means most
of the solutions which are in the market today will fail to survive.
So another headache for users is this. If you're using one of the 50
to 100 or so SSD software solutions which are in the market today... how do you
insulate yourself from the pain of having picked the
wrong product - when
those solutions stop being supported in new SSDs in 2 to 3 years time?
In
one way - the need to architect IT infrastructure for the certainty that
everything will change due to the growing disruptive impact of SSDs is similar
to the idea of designing earthquake-proof buildings. The difference being
that everywhere is going to be an SSD zone - no one can be sure yet how to
make their own infrastructure SSD future proof. | | |
| ... |
 |
| ... |
the Top SSD Companies
SSD endurance -
forever war
can you
trust SSD market data?
how fast can your SSD
run backwards?
the top 100 SSD
articles on StorageSearch.com
meet Ken - and the
enterprise SSD software event horizon |
| ... |
See also:-
100 popular SSD
articles on StorageSearch.com
StorageSearch.com is published by
ACSL | |
|
