reporting the SPARC® market since 1992
| Given the
lacklustre content of many Sun Microsystems
press releases during the
last 18 months, I started speculating idly what sort of announcement the company
would have to make, to create the kind of enthusiasm in customers and market
analysts that they used to get in their white hot days. Hey, if we're in make
believe mode, we might as well go for the gold. It would have to be a product
announcement that marked the inflexion point in Sun's revenue slide, and the
start of a strong steady rise in revenue, and real server market share (not just
Solaris server market share).
How about some good news about its X86 Linux servers?
...No. I dismissed that straight away.
Even if Sun quadrupled its X86 Linux server revenue, it would still be getting only about 2% of the Linux market.
How about some good news about Java?
...No, the Java hype machine has lost its power to buy votes where it counts - in increased server sales.
Let's look at the best case. The best news for Sun's investor would be if Sun were to announce it was spinning off its Java business as an independent separate company. It's a useful language, and a worthy successor to vanilla C. But you don't have to own a cow to drink milk. Sun's ownership of Java hasn't guaranteed strong server sales recently. But getting rid of the expense and distraction of supporting Java won't turn around Sun's fortunes either. In fact, if Sun spun out its Java unit and promised to support all available web software technologies including those from Microsoft, Sun's share price might even go up - for about 15 minutes until everyone realised that they didn't really mean it. No, overall, Java is not the Trigger Event. Not even if Microsoft did a deal with DoJ and was able to acquire the Java business for a generous sum of money. Ownership or otherwise of Java is probably revenue neutral. We have to look elsewhere.
How about some faster SPARC processors?
...This won't be the trigger event either.
The only reason Sun still has lots of customers is because everyone expects and knows that Sun will come out with faster SPARC processors - eventually. They've been behind the curve for over a year, so there's a good chance that soon we'll see some leapfrog products soon. But this is what marketers call a "hygiene factor" - something you have to do, or else! But not intrinsically a factor which differentiates you from competitors. Faster SPARC processors will slow down Sun's decline, but will not be enough, by themselves to trigger server revenue growth. Everyone else from Intel to IBM is also in the faster processor business. It's a race in which Sun has to participate, but there are no winners. When you reach the end of the track, you have to start all over again in the next race, in which you are expected to run even faster.
So what would it take?
There is a solution, and it's deceptively simple.
Think about how the market would react if Sun were to launch a complete range of supported SPARC servers running Linux.
It would send out two strong clear messages...
1- SPARC can compete with Itanium in the open Linux market, and beat it.
2 - SPARC/Solaris customers can be confident that their hardware platform has a long future, and will remain competitively priced.
You can already get Linux on SPARC from 3rd parties, and you can get Linux from Sun. But if you look at Sun's own Linux page you won't see any mention of Sun selling a Linux for SPARC.
Why not now? is very simple. SPARC processors aren't currently fast enough. A SPARC Linux server with today's chip technology from Sun would be a dead duck. More expensive and slower than Itanium.
But it doesn't have to be that way. The next time that Sun announces a new SPARC chip family it could use the opportunity to make a big splash in the Linux pond.
With the confidence of fast enough SPARC processors,behind it, and chip pricing which anticipated the higher volumes that would come if Sun were making a credible stab at grabbing 15% of the Linux market, Sun would be a serious contender. Sun would also have to convince a bunch of other OEMs to license SPARC technology, but they've done that before. If the prize were the fast growing Linux market, instead of the slowly declining Solaris market, OEMs will be queuing up in the aisles to use SPARC technology.
Sun has dismissed this option in the past because they were worried about the cannibalization of their SPARC/Solaris customer base. In my view that was muddled thinking. Customers who today migrate from SPARC/Solaris to other architecture Linux will do that anyway.
The future availability of a strong SPARC/Linux product family will be beneficial to Solaris users - there will be a bigger market for processors, and so faster SPARC processors will come out sooner and at lower cost. It will cut down the arguments for users to migrate away from SPARC. It will also offer a credible platform for Sun to attack HP and IBM. It will be good for Sun, good for current and future SPARC users, and good for competition in the computer market.
So why doesn't Sun just do it?
Surviving the Solaris x86 Wars
the Solid State Disks Buyers Guide
a Short History of Disk to Disk Backup
the Fastest Growing Storage Companies
RAM SSDs versus Flash SSDs - which is Best?
|...Later:- 3 years
later - it looked like Sun was going to follow this prescription.|
Sun Embarks on the Business Recovery Plan We Advocated in 2003 - Linux on SPARC
Editor:- February 16, 2006 - Sun is hyping up the prospects of a Linux port to its UltraSPARC T1 processor.
To facilitate this - Sun has published details of the SPARC architecture and relevant APIs for free download.
Commenting on this Jonathan Schwartz, President and COO, Sun Microsystems said "Today we open the door to expanding SPARC onto new platforms and into new markets, breaking down barriers to innovation and giving our customers more choice. Having Linux or BSD ports for the UltraSPARC T1 processor will greatly expand the SPARC market, giving customers more opportunities to reap the benefits of our CoolThreads technology. The OpenSPARC effort is fostering a community for SPARC-based, 32 thread innovation that will play a crucial role in redefining industry standards in the data center."
5 years later:- we saw this announcement
SPARC T2 Gets Carrier Grade Linux
SAN JOSE, CA - April 16, 2008 - Wind River Systems, Inc. today announced it will port its Carrier Grade Linux and Workbench development suite to Sun's UltraSPARC T2 processor.
This will be the first carrier grade Linux for Sun's CMT processors. Sun's Netra Carrier Grade rack servers and ATCA blades will be the first CMT systems to run Wind River Carrier Grade Linux. ...Wind River profile, Operating Systems for SPARC servers
Editor's comments:- in SPARC's 20 year history Linux software hasn't had much close contact with SPARC hardware - and mainly featured as a competitor.
Despite that, I was surprised to see that the word "Linux" appears on 24% of pages in the SPARC Product Directory.
That's more than I thought but probably 4x less often than in other "Unix" publications.
In August 2003 (which was not a very optimistic time for the SPARC market) I wrote an article - What's the Trigger Event that will Turn Around Sun's Revenue Decline? - in which I explored all the technology and business options that could make a significant difference to how the market viewed and reacted to Sun's SPARC products.
The analysis in that article is just as relevant now and still makes good reading. And I haven't changed a single word in this conclusion from that 5 years old article.
"The future availability of a strong SPARC/Linux product family will be beneficial to Solaris users - there will be a bigger market for processors, and so faster SPARC processors will come out sooner and at lower cost. It will cut down the arguments for users to migrate away from SPARC. It will also offer a credible platform for Sun to attack HP and IBM. It will be good for Sun, good for current and future SPARC users, and good for competition in the computer market."
3 years after that article was published HP wrote an (anti-Sun) article - The Real Story about Linux on Sun's SPARC - with some amusing and interesting points about what they referred to as Sun's "on again, off again" approach to Linux. That, and the stats quoted by HP may all be true, but I think the new combination of SPARC T2 with Linux is going to reduce the bonuses paid to many HP server sales people in the next year or so.
|today's SPARC news||SPARC computers||SBus & PCI cards|
|SPARC manufacturers||Sun/SPARC Resellers in the USA||Sun/SPARC Resellers in the UK|
SPARC(R) is a registered trademark of SPARC International, Inc. SPARC PRODUCT DIRECTORY(SM) is a service mark of SPARC International, Inc used under license by ACSL. Products using the SPARC trademarks are based on an architecture developed by Sun Microsystems, Inc.