|
by Zsolt Kerekes,
editor - February 3,
2010 |
The
history of
Sun Microsystems'
SPARC/Solaris business can be broadly divided into 2 parts.
1 - the
Growth Years - 1987 to 2000, and
2 - the Decline and Lonely Wilderness
Years - 2001 to 2009
Sun's biggest strategic success in the Growth
Years - was that the company designed a processor family (SPARC) - which at
one stage (in the mid 1990s) attracted more than 6 competing,
compatible chipmakers who supplied SPARC CPUs for use in workstations
and servers that competed with Sun itself. Just in case you thought this kind
of thing only happened in the
Intel
x86 world (Intel, Cyrix, AMD etc).
Apart from Intel Architecture -
SPARC is the only processor chip family which has achieved this level of support
within the mainstream commercial computer market. Sun, which was a small company
when it started, used clever partner marketing programs to create the illusion
that it was building an open systems server market. More than 50 computer makers
and hundreds of bus compatible card makers helped Sun sustain this dream of
creating an open server architecture - which would be faster and more reliable
than Wintel, while being cheaper and safer to buy than "proprietary"
products from IBM, DEC and HP.
2001 - Sun loses more than revenue!
2001
marked the turning point in the 2 phases of Sun's history.
In 2001 -
all server companies were suffering from the dotcom recession - which was
extended by the events of 9/11. But in this year Sun lost more than a share of
diminished user IT budgets (which would increase later for its main rivals.)
Sun lost its reputation for reliability. For more about this
see the article - Unsafe At
Any Speed? - a contemporary exposé of an easily avoidable design
mistake which affected tens of thousands of high end SPARC servers . That was a
pivotal point for most Sun customers. If they couldn't trust Sun, and Sun's
SPARC servers were slower and more expensive than alternatives - then why
were users risking their own operations by relying on this flaky outfit?
That's when many die-hard SPARC user sites decided that the bitter taste of
migration
planning was better than the risk of being poisoned.
Sun's biggest strategic failure in the Decline and Lonely
Wilderness Years lay in its many failed attempts to achieve technology
leadership and relevance in the
storage market - and
most importantly its failure to recognize the threats and opportunities for all
server makers (not just Sun) posed by
SSDs.
Once upon a
time - in the early years of the new millennium - as Sun started to lose its
earlier performance advantages (because Microsoft was supporting Intel's server
chips much sooner - instead of ignoring new features for years like they had
done before) and as all server CPU designers were looking ahead at slower
improvements in clock speeds - Sun had unique reasons and opportunities for
exploiting SSD accelerated architectures. I wrote an exasperated wake-up call
to Sun in my (2004) article
Why Sun Should
Acquire an SSD Company. They controlled the OS, they controlled the CPU
design and they sold servers. What a fantastic starting point for the next phase
in computer architecture!
Instead Sun did virtually nothing to
redress the problems caused by lagging CPU clock rates - and
mindlessly pursued the easier option of fatter - rather than faster -
processors.
This was a market strategy which solved some problems in
some markets for a short time - but could not be sustained as the core SPARC
base was declining - because other chips were much faster. Smaller revenues from
SPARC servers - meant smaller investments and slower design improvements in
SPARC chips. Sun's real strength had always been integrating and supporting new
chips and interfaces with its OS. It wasn't the best at designing chips.
Will
Oracle do any better?
Oracle has bought a server company at a time
where the server is becoming almost irrelevant.
If you look at a
datacenter in 2015 and look at where the money will be spent - the proportion
spent on servers will be less - and the proportion spent on solid state storage
will be much more than that spent on the processors, motherboards and RAM.
The
key value for Oracle in Sun is its OS.
If Oracle can tweak Solaris so
that it can automatically tune the best performance out of the attached SSDs -
it would have a great product to pitch.
But the autotuning SSD
acceleration market has already started. The race for the
SSD ASAPs market is
application agnostic.
The computer market is rushing ahead to a new
bubble - and it will
take another 5 to 10
years before the winners are announced.
Want to know what's going to
happen?
Retune to StorageSearch.com
- which since 1998 - has been "leading the way to the new enterprise
storage frontier." | |
|
|
| . |
| |
. |
 |
|
|
|
 |
| . |
| this way to the Petabyte
SSD |
In 2016 there will be
just 7 types of SSD in
the datacenter.
One of them doesn't exist yet - the bulk storage
SSD.
It will replace the last remaining strongholds of
hard drives in the
datacenter due to its unique combination of characteristics, low running costs
and operational advantages. |
 |
... |
The new model of the
datacenter - how we get from here to there - and the technical problems which
will need to be solved - are just some of the ideas explored in this
visionary article. | | | |
| . |
| the Problem with
Write IOPS |
the "play it again Sam"
syndrome
Flash SSD "random write IOPS" are now similar
to "read IOPS" in many of the
fastest SSDs.
So
why are they such a poor predictor of application performance?
And
why are some users still buying
RAM SSDs which cost
more than SLC and significantly more than MLC? - even when the IOPS specs look
similar. |
 |
This article tells you
why the specs got faster - but the applications didn't. And why competing SSDs
with apparently identical benchmark results can perform completely
differently. ...read
the article | | | |
| . |
| |
