What do data recovery stats
tell us about HDD vs SSD reliability?
Editor:- November 7, 2010 -
how do SSDs compare with
HDDs for
reliability in the
real world?
Forget all the theoretical calculations about MTBF - which are
grossly over optimistic...
In a short article in the right hand column
of this page I share what's known about the stats for SSD data recovery. See
the footnotes to the article about ActionFront - to learn more.
News to me - Seagate has MRAM technology
Editor:-
November 5, 2010 - an interesting article on Denali's blog site
discusses
Seagate's
relationship with MRAM.
What's
MRAM?
- It's a non volative memory
technology which is a long odds contender for use in some parts of
SSDs.
Memoright fills 2.5" SSD performance gap
Editor:-
November 3, 2010 - Memoright
has
launched
its first range of MLC based SSDs - the
2.5"
FTM-25 range are SATA
2 compatible - and have upto 400GB capacity.
Editor's
comments:- from the performance figures (50k / 30k R/W
IOPS) I
guess they use a SandForce
controller. But as
I've said before there's a lot more in the price and makeup of an
SSD than
just the controller
and flash chips.
Memoright's newest SSDs have over-voltage protection and over-current
protection and their quality management system (evolved for military SSDs)
offers 100% traceability for every SSD.
3 years
ago (Oct 2007) - Memoright was shipping the fastest
2.5" flash SSDs.
Then for a few years the company seemed to be lagging behind in the speed
stakes. The new products will appeal to designers who need industry standard
speed with some extra reliability
features added.
Wanted! - Mr Right (partner) for STEC
Editor:-
November 2, 2010 -
STEC today
announced
that revenue for the quarter ended September 30, 2010 declined 12%
from the year ago period to $86 million) although it was 40% higher than the
preceding quarter.
Editor's comments:- I've said this many
times before in past years - STEC has weak routes to market - like
EMC (ranked #20 in the
top 20 SSD companies).
STEC still has a stronger brand in the SSD market than any of the oems it
partners with. It must dump them, bypass them or run the risk of fading away. In
the enterprise
and military markets
in which STEC engages - it should be reporting 50% to 100% year on year revenue
growth - not double digit revenue declines.
If I was looking for an
analogy here - it would be that STEC decided several years ago on a safe "father
figure" route to success in the mainstream market. Instead it should have
gone for the James Dean
or Elvis
approach. SSDs are a disruptive market - a message not best delivered by men in
suits. EMC is no
Darcy
either.
Foremay ships fastest 1.8" SSDs
Editor:-
November 2, 2010 - Foremay
today
announced
volume shipments of the
fastest
1.8" micro SATA slim
flash SSDs - with 280MB/s R/W and random
IOPS as
follows:- up to 30,000 read and 15,000 write.
The 5mm high SSDs
have capacity up to 400GB and are available in industrial temperature versions.
Editor's
comments:- I asked Foremay - who designed the
SSD controller?
I
was told - "It is joint solution between Foremay and a 3rd party" -
whose identity they are keeping to themselves.
Walton Chaintech blends controllers in new consumer SSD ranges
Editor:-
November 2, 2010 - Walton
Chaintech today
announced
availability of 3 new 2.5"
SATA 2 SSD ranges for consumers and gamers which feature
SSD controllers from
SandForce (skinny SSDs -
from $147 60GB) and Indilinx
(regular SSDs
- from $91 30GB).
Demartek tests LSI's CacheCade
Editor:- November 2,
2010 - Demartek
has published a sponsored
test
report (pdf) which compares the performance of
SSDs and
HDDs in a simulated
web server environment when managed by LSI's
CacheCade
software - which provides
SSD ASAP
functionality.
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
off.
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?
How fast is the SSD market changing?
Editor:-
November 1, 2010 - if you look at the top 20 SSD topics viewed by StorageSearch.com readers in
October - (listed on the market
research page) they look very similar to the same list a year ago.
Is
this a sign of the SSD market maturing?
It's tempting to think so - but
that would be a mistake in my view.
Yes there is traction. The main
concerns in the SSD market today are very similar to what they were a year ago
- but there are new ways of addressing those concerns too.
And a lot
more SSD vendors who want to persuade you that their own particular way is the
best. You can see how the popularity of individual SSD companies has changed in
the past 3 years by viewing the top 10 SSD oems list in the 3rd quarter in these
years -
2010,
2009 and
2008.
When
you look at daily or monthly changes in the SSD market - the cumulative changes
don't give a clear picture - because the sampling interval is too small. But
enlarge that window - and it's easier to discern the start of momentous changes.
Here
are some segments in the SSD market which were invisible or didn't exist a few
years ago - but are worth thinking about seriously now.
2 years ago -
the
PCIe SSD card
business was just starting and didn't register in any top 10 lists. Today it's
shaking up the enterprise
SSD market.
2 years ago - outside of a small technical elite - most
customers didn't ask the question - who had designed the
SSD controller
in the SSD they were buying. If they thought about it at all - they assumed it
was the same company whose logo was on the invoice. Today the
controller identity
has become one of the key ways that vendors differentiate their
SSD brands.
2
years ago - the SSD
ASAP market didn't exist. It's still a very small market today - even
though 10 to 20 vendors have shipped or announced ASAP products in the past
year. That's because
users are
being very cautious and evaluating these systems - rather than taking on
trust vendor claims that they can accelerate pools of HDD storage automatically
and economically. If vendors can convince users that their ASAPs work - and
are a low risk way to accelerate legacy apps and infrastructure - this could
become a multi-billion dollar market.
2 years ago - the SSD backup market didn't
exist. Today it's still a tiny niche - because it doesn't offer economic or
reliable solutions for the mainstream. That's a slow burner market - with many
technologies still to be developed and prove. My most optimistic forecast
indicates it could be another
5 years before it
this one become mainstream. It will still use
2.5" SSDs - but
they will be unlike any products in the market today - optimized for
reliability, low power consumption and data integrity rather than speed.
2
years is a short time in the
history of
the SSD market. But as you can see from my examples above - if you simply
assume that the trajectory of the SSD market will be like that of speeded up
hard drives - you'd be wrong about where the market's destination. |
|
|
......................................................................................................................... | |
|
. |
What do data recovery stats
tell us about HDD vs SSD comparative reliability? |
Nowadays - in 2010 - there has been a lot
of speculation in the storage market about the
reliability of flash
SSDs in the field - as opposed to their theoretically calculated lifetimes
due to factors like write intensive
wear-out.
Having said that - the MTBF figures quoted in some company's
datasheets become fiction for
flaky SSDs - which
fail at the start of their life because the design was wrong - and which need
euphemistically called "firmware upgrades" to make them work properly.
How
can you get an undistorted industry wide view of the SSD reliability
experience?
SSD data recovery
is a subject I've discussed in detail with many leading DR companies.
In this context I mean trying to recover data from an SSD in situations where
there is no useful backup.
The main causes of SSD failure are power
surges, static, environmental stress, software anomalies and ineffective flash
wear leveling.
Extrapolating from the data I've seen suggests that the industry is
handling in the low hundreds of SSD recoveries per month. That compares to an
installed base of around 40 million SSDs.
What can you infer
from this?
If
hard disks were as
reliable as SSDs - the data recovery business would be much smaller than it is
today.
SSD data recovery is a tiny business today.
It's so
small that it's not economically viable for any DR companies to advertise
their SSD recovery services here on StorageSearch.com.
That may be
bad news for SSD market publishers - but it's good news for prospective SSD
buyers. The message from the data recovery market experience is - that if you
choose the correct type of SSD for your application - it will be much more
reliable than any hard drive. | | |
. |
|
. |
|
SSD jargon - explained |
Do you understand all the clues
in the solid state drive headlines? This article will help. | |
. |
|
. |
SSDs - the big picture |
Editor:-
StorageSearch.com was the
world's 1st publication to provide continuous editorial coverage and analysis
of SSDs (in 1998) and in the 12 years which have followed we've led the market
through many interesting and confusing times. |
|
If you often find yourself
explaining to your VC, lawyer or non technical BBQ guests why you spend so
much time immersed in SSD web pages - and need a single, simple, non very
technical reference to suggest - this may be the link they need. | | | |
. |
|
. |
| |