it Black ? or Not
Fade Away? |
Sun's new songs play better for a new generation of
SPARC Product Directory
- March 29, 2005
Active marketing support for Sun's SPARC
platform by Independent Hardware Vendors in the first quarter of 2005 has
declined to a lower level now than it was back in 1990, the year after Sun
launched the SPARCstation 1. Does this mean a bleak prospect for users and
stakeholders in the market? This article, based on first hand contact with
companies in the market, describes what's happening now in the SPARC
compatible systems market , and puts it into a historical context. It also
predicts which segments in the Sun compatible hardware market have the most
upside potential for suitable partners.
All markets have their ups and
downs. But the Sun compatible SPARC systems market has reached a stage where
it's hard for most observers to understand what's happening. As the market has
changed from being a pseudo open market to a de-facto proprietary one - most of
the information about what's really going on is not plainly on view. And what is
revealed (by Sun) can't always be relied on - due to optimistic hype, deliberate
competitive misinformation, and poor competitive analysis by Sun's own
So how can SPARC users, marketers, technology partners and
competitors make sense of what's going on - in a fashion which helps them form
effective strategic plans?
Well just as the CIA's Kremlin watchers
in the cold war - could pick out threads from clues which most others would
miss and from them predict the most likely directions of change - we can do
the same. As a long term Kremlin and SPARC market watcher, I will try to bring
to your attention some useful currents. Some are going downstream, some are
going upstream and some are stalled. It's a mixed picture.
Is the SPARC
market growing in server shipments? - Yes - most likely.
Is the SPARC
market growing in year on year revenue? - Yes - most likely. But it's running
at about half the peak revenue level of 5 years ago. It's not likely that it
will regain that high ground within a forseeable timeframe. But it's not
impossible either. A recovery to historic high revenue levels would rely on
Sun's server competition having some very serious upsets - like a faulty
processor - or a series of bad operating system problems. (Which have all
happened before to various companies in the IT market.)
One thing that
is clear to anyone looking at the signs is that the 3rd great cull of SPARC
compatible manufacturers is currently in progress and well underway. This
competitive extinction of companies and products from the SPARC landscape is
quite different to those which have gone before. It's creating less waves. But
the impact on the market will be quite profound. What's not so clear, is that
there are many parts of the SPARC systems market today which are positively
bubbling with high energy growth and enthusiasm.
Let's look at some
market history to see how we got to this complex brew.
The 1st Great Cull of SPARC Compatible OEMs happened in the
period 1997 to 2000.
This was triggered by two technology changes
introduced in Sun's SPARC systems in 1995 to 1996
- the introduction of 64 bit SPARC processors, and
The transition to 64
bit SPARC processors was a step which triggered the market exit of the leading
32 bit SPARC chip company Ross Technology whose hyperSPARC chip had first
competed with Sun, and then ended up being used by Sun itself. Although Fujitsu
also launched their own 64 bit SPARC chip at the same time as Sun (well
actually a few weeks before) Fujitsu's strategy was to use this in their own
systems - and they did not supply the SPARC64 chip to other makers of SPARC
systems. At that time (Q4 1995) there were dozens of manufacturers of SPARC
systems. Many struggled to compete with Sun for a year or so using faster 32
bit chips. But the performance race at that time went to the 64 bit SPARC chips.
(Unlike the performance race in 2001 to 2004 in which 32 bit architecture
processors typically outperformed 64 bit chips - because their higher Gigahertz
clock rates compensated for their narrower register I/O datapaths.) So, back in
the late 1990s, SPARC systems makers had to turn to their main competitor Sun
to buy 64 bit processors or get out of the business. Sun did a good trade
supplying its own SPARC processors and motherboards to its system level
competitors upto about 1999.
- the adoption by Sun of the industry standard PCI bus to replaces Sun's
own SBus - which had been introduced in 1989.
Sun's switch to the PCI bus in 1996
lowered the entry costs for host bus adapter manufacturers who wanted to sell
their products to Sun users. Instead of having to devote large resources to
understanding and supporting Sun's SBus, HBA makers could just write Solaris
drivers for cards which they were already planning to sell in the PC and Intel
Architecture server markets. This opened up genuine free market competition.
Many of the companies which had focused exclusively on the Sun market were
cushioned from the transition from SBus to PCI for a year or so by supplying
the aftermarket, but in the sweet spot of the market - in supplying HBAs for
Sun's newest workstations and servers - the niche SBus makers now found they
were competing with bigger companies which had more distribution channels, lower
manufacturing costs and better marketing organisations. Another factor towards
the end of the 1990s was that the market share wars in technologies like
Fibre-channel adapters, SCSI RAID controllers and telecoms adapters were coming
to a close. The consolidation and reduction in HBA makers which resulted at
this stage was due to the natural workings of the market. Sun and its users got
the benefits of lower adapter card costs and access to a bigger technology base
than previously available via SBus.
The 2nd Great Cull of SPARC Compatible OEMs happened in
In the second half of 2000 the dotcom bubble started to burst.
This impacted the Intel PC market first - but its effect on Sun stayed hidden
until April 2001
when Sun reported its financial results for the first calendar quarter of 2000.
Then later in the year - the terrorist atrocities of 9/11 created a business
climate that put the seal on what would eventually become a 3 year IT
Although most of the IT market was negatively effected by
these events - there were specific actions by Sun which made things worse for
those SPARC compatible OEMs whose products used SPARC chips or motherboards
supplied by Sun.
It was reported to SPARC Product Directory by many
vendors that Sun substantially changed the way it did business with them.
Sun allegedly increased the prices of its SPARC processors and
motherboards which ran counter to the declining price trends in the rest of
the IT market, and therefore made it impossible for most of its previous OEM
customers to continue integrating these products and selling them at a profit.
While Sun's market had been growing it had been happy to have other IHVs
developing the SPARC customer base. But faced with a static or (as it turned
out) steeply declining market, and with a large amount of virtually unused SPARC
systems coming back onto the market from liquidated dotcoms, SPARC Product
Directory believes that Sun used its monopoly position in the supply of SPARC
chips to snuff out anyone who could be classified as a head on competitor.
The bitterness which many of Sun's former technology partners felt
towards Sun's predatory practises at that time ensures that they will go out of
their way to avoid future partnerships which involve Sun or its technologies. So
I think that Sun's attempt to position Solaris 10 as an open source operating
system for Intel Architecture servers will not attract back many lapsed SPARC
The 3rd Great Cull of SPARC Compatible OEMs - Q4 2004 to
The IT market has been growing for more than a year. But the
recovery in the Sun market lagged the general server and PC recovery by 2 or 3
quarters. Many of the SPARC OEMs who survived the recession have been quietly
dropping their support for the SPARC platform. To paraphrase many of the
conversations I've had on this subject recently.
"The Sun market
"The Sun market is not growing."
see much better opportunities for growth outside the Sun market."
outside - many of these companies look pretty much the same as they did before.
But there has been less talk of SPARC. The budgets for promoting SPARC products
have been terminated. The product catalogs will still show the same old SPARC
products that were there before - for a while. From the inside - according to
those still inside and many who have jumped ship to competitors - the priorities
"The product roadmap for SPARC ended with the last
That means users will see much less choice in suppliers
than they saw before.
DRAM memory market grew 25%
in revenue in 2004 - so you'd think that 3rd party memory vendors who sell into
the Sun market would be happy. But I've heard reports from both VARs and OEMs
suggesting that they will pull out of the Sun memory market. The reason? Sun has
been offering very competitive pricing on Sun branded memory in the last year -
often cheaper or little more expensive than 3rd party memory.
Some SPARC IHVs Have Been Reporting Significant Growth -
in 2004 / 2005
As I said at the start of this article. The Sun SPARC
compatible hardware market presents a complicated picture. I've discussed the
negative trends above but there are a number of areas in which SPARC IHVs have
been doing well or in which the vendors are mostly upbeat and confident about
current and future business.
is a product area which has seen an explosion of new products and suppliers in
recent years. SPARC notebooks have been getting faster, with dual processor
notebooks now available with performances as fast (or sometimes faster than dual
processor rackmount servers).
One customer of
was surprised to find that their Tadpole notebook, when used as a network
server, ran faster than their Sun server which used the same processor.
After a technical investigation the reason turned out to be that because the
notebook was designed with a small motherboard - the design ran the memory with
less wait states. The result was the opposite of what you'd expect.
The main reason that PC notebooks are sluggish is that they use low
power disk drives which run at 4,200 RPM compared to 7,200 RPM for many desktops
and 15,000 RPM for servers. This creates the impression that all notebooks are
sluggards. But if you attach a fast disk to a well designed SPARC notebook you
can get a pleasant surprise. SPARC notebook makers have tried for years to get
users to think of them as disaster recovery machines. You can lock one in a
fireproof safe and have it ready to restart your enterprise (maybe with less
users) if you have a catastrophic failure of your main server. Which can just
as easily be caused by an
error as hardware failure.
SPARC notebooks have also been
getting cheaper and lighter at the low performance end. Why would you want a
low performance SPARC notebook? Well - they have a much lower cost of
ownership than desktop machines when it comes to running training courses on
customer sites. And they are a useful sales and development tool for software
companies who support SPARC / Solaris software distributions.
the notebook market done so well for SPARC IHVs? There are 3 reasons:-
- The notebook market generally (not just SPARC) escaped the worst effects of
the IT recession which hit desktops and servers. It was fueled by the growth of
a more mobile workforce who use notebooks instead of desktops even when they're
in the office.
- SPARC notebook makers were not targeted by Sun's death rays. Sun flirted
with making its own SPARC notebooks in the early 1990s - but gave up after
making some dud products - and left this market to specialist manufacturers. So
the notebook sector escaped the blasts of internecine warfare which killed off
all the commercial SPARC server makers which used Sun's chips and motherboards.
In fact Sun is a reseller of some SPARC notebooks - and that's been a big boost
growth area in the SPARC systems market in the last 4 years has been in
military, intelligence, homeland security and defense applications.
- Increased demand by military customers (see below.)
seems unbelievable now, that when I spoke to many
manufacturers at the start of President George W Bush's first term in office,
that concerns were raised that the new administration would cut back on defense
spending as part of their avowed plans to reduce big government and reduce
taxes. All that was changed by the events of 9/11. The increased spending in
the military and defense markets after 9/11 and in the build up to the 2nd Gulf
War provided big opportunities for all suppliers in the SPARC / Solaris market.
The widespread public concerns about viruses and security leaks in Microsoft
platforms meant that the SPARC platform, which otherwise might have been reduced
in scope at this time due to performance parity by Intel Architecture
processors, instead became an important strategic tool which was widely
regarded as being less vulnerable to electronic attacks.
manufacturers exit the SPARC market, the updraft from their departure
creates business opportunities for those who remain committed to the SPARC
In the market for blade products or SPARC
SBCs - the number of active suppliers (apart from Sun) has dwindled from 8 (6
years ago) to just one -
You don't need to be a business guru to see the growth opportunities that
creates. In fact Themis is expected to benefit greatly from the apparent
recent exit in 2004/5 from the SPARC market by the last of their long term
rivals Force Computers and Continuous Computing who appear to be focusing
instead on their Intel Architecture product lines. In the last few months the
company has made a series of announcements about opening new sales offices and
new sales hires - presumably to provide more feet on the street to welcome on
board long time SPARC users who have been left stranded by the departed
The storage market has
been an expanding area recently for SPARC IHVs.
In a series of
articles published in
2000 I explained why Sun's storage policy at that time, which was based on
and internally developed products which failed to meet the needs of most of
computer users - who weren't using Solaris - was doomed to failure. In the years
that followed - this was confirmed as market data suggested that nearly all
Sun's storage sales were in the captive Sun server market, and that Sun's market
share outside the Solaris/ Linux base was almost nil.
All that has been
During 2004 Sun confirmed oem and reseller
agreements with over a dozen strategic
Sun's new strategy is to get the storage business (which has become a
bigger part of the
datacenter IT budget than servers) by being a distributor of best of breed
storage products instead of designing these systems itself. In many cases the
products are branded as Sun products - but are not reengineered - but merely
rebadged versions of commercially successful products. Storage OEMs have been
falling over themselves to be included in Sun's storage boutique.
strategy, which worked for Compaq
before its merger with HP in 2002, represents a business which is currently
worth around $2 billion per annum for Sun (including internal and external disk
and tape storage). If Sun can leverage its strong Unix credibility and
marketing machine better within the wider Linux server market to target
customers who buy other brands of Linux server - then it's possible that in the
period 2005 to 2007 - Sun could earn significantly more revenue from its
storage reseller business than from its AMD server business. In fact
growth in Sun's storage business in the next few years could realistically
take the company to $3 billion storage revenue.
...Later note:- this
article was published in March 2005 before Sun's (June 2005) announced
acquisition of StorageTek. The combination of that and Sun's acquisition of
Procomm's NAS technology (May 2005) means that Sun's storage related revenues in
2005 could be on the order of $5 billion. If successful - this could make Sun
one of the 10 biggest
storage companies in 2008.
Unlike the old Sun compatible server
market in the 1990s when most SPARC IHVs regarded Sun as head to head
competitors, today most of the storage OEMs I've spoken to have been
enthusiastic about Sun's adoption and remarketing of their products. They see
Sun opening up big markets which they would be unable to reach through their
own efforts. Upcoming changes in the storage market in 2005 like
Serial Attached SCSI,
Sun can influence by its server side technologies, will create new lock-in
opportunities for Sun sales people as users will initially buy those products
from server companies until they are comfortable with 3rd party compatibility
So, overall, I expect we will see a surge of new
manufacturers who want to sing in Sun's new band.
|more articles by the same author:-|
the Top 20 SSD Companies
this way to the petabyte
introduction to enterprise SSD silos