Sun Microsystems is focusing on Opterons and Solaris x86 for a good reason. Sun has figured out a way to make x86 hardware cheaper than anyone else. Sun is hoping that by leveraging its cost advantages the company will be able to ensure the success of its x86 hardware and Solaris.
Sun Microsystems’ x86 Strategy
About The Author
Adam Scheinberg
Technology Executive • Web Developer • Father • Foodie • Music Snob • OS enthusiast
Follow me on Mastodon @[email protected]
I don’t know if this is a tad naive, but it seems to me that having some open source competition for Linux should be a good thing for Linux, something that most people seem to miss when they’re busy slagging off Sun and Solaris. I don’t think Sun will fight ‘dirty’ against Linux (in the same way as Microsoft might, embrace, extrend, distroy etc), and even if they tried this i don’t think they’s get away with it.
Cheers!
–Rob
Which is very good. The fact that the few bucks you spend on Solaris when buying it goes to R&D unlike investments in Red Hat is a good thing to me and is definitely well spent money. I can’t wait to get my hands on ZFS =)))
Hopefully the MS blessing will mean that Solaris/Windows networks will work even better in the future…
Go Sun Go
Richard Pryor, when describing George Foreman’s boxing style, he said the following:
George Foreman has a unique boxing style, NONE!
So, with that said and looking at Sun Microsystems as of late, their execs can’t seem to figure what they’re doing. You have Schwartz who doesn’t have a faintest clue of what garbage he spews out each time he opens his mouth, you have their CEO, McNealy making these alliances with Microsoft where he knows in the end they’re going to end up screwing him(Sun) when the smoke clears.
So you tell me, what is their strategy?!?!?!
The answer: NONE!
So you tell me, what is their strategy?!?!?!
The answer: NONE!
THat’s funny, I see Linux keep growing on the market, and I thought Linus just said that he has no strategy, he pokes a little here and there…. must be a winning concept has then aye?
Sun isn’t abandoning Sparc. They are spending money overhauling it to the next generation. Sparc will play a big role in Sun’s future. Sun was one of the first companies to see that the mhz game is over and they changed directions pretty immediately. I remember Sun was talking about releasing their 32 core cpu’s about the same time Intel was talking about releasing 4 ghz cpu’s. Sun will be among the first companies to have a nextgen chipset. The Sparc legacy is far from dead.
Sun is serious indeed about Opteron. The fact that Solaris 10 is _single_ source code for Opteron, Sparc, and X86 show that Sun no longer treats Opteron and X86 (IA32) as 2nd class citizen.
The Sparc info is wrong. Niagara and US IV+ were shared at recent MicroProcessor forum or HotChip.
With 32 threads, nothing will come even close (in terms of threads) when it is released in year 2006. The current leader, Power5, has only 4 threads.
Opteron is a perfect fit for Sun’s low to mid-range servers. Opteron is better than Xeon, Opteron isn’t Intel, Opteron isn’t awfully expensive, geeks like Opteron, it runs Linux, it runs Solaris, it runs Windows, what isn’t there for Sun to love about it?
Sun is also continuting SPARC through its next-gen multi-threaded processors (e.g., Niagara). SPARC will always be the CPU for gigantic SMP, as Opteron is optimized for a different market. There are also reliability features in SPARC that other CPUs lack. So, SPARC is here to stay for a long while.
Don’t forget that Sun’s new pricing model is totally competitive with RedHat all the way from free up to full enterprise support. Solaris 10 is set to go head-to-head with RedHat in the enterprise market. That’s good, because Solaris is better than RedHat in most ways, such as documentation, configuration stability, maintainability (this is not a troll, because I really like Linux, just not RedHat–I’m more of a Debian/Slackware person…and *BSD).
with Sol 10 is the almost comically limited driver support. I’ve tried with a Dell PowerEdge 2650, PowerEdge 2600, and IBM xServer 206 — never got a complete system operational. On each system a different hardware component wouldnt work, network, video, SATA controller….
I know that hardware support is the thing that is most difficult for any OS, but Sun needs to improve their offering dramatically if they expect any kind of mass exodus from linux.
Don’t get me wrong, I think the features of Sol 10 are very impressive, but in the end if my only option is to purchase hardware from Sun, then not much has changed.
I don’t know but I have Solaris x86 working in a production or tests on IBM x335, IBM x345, Dell 2650, Dell 6650 and more… Actually HW suport for x86 servers is quite good.
This is one area where Sun can benefit from other open source projects. It should be much easier to port an open source Linux (or better yet, FreeBSD) driver to Solaris than it would be to write one from scratch.
in the end if my only option is to purchase hardware from Sun
Well, their prices are competitive with Dell (apples-to-apples comparisons), their x86 servers are RedHat and Windows certified, so you really aren’t losing anything by buying Sun. Their value-added above Dell would be Solaris 10, their support options, and a purple decor for your server room.
Also, Sun is _not_ a Microsoft reseller like Dell is, so they have much greater flexibility in what they can sell you.
> So you tell me, what is their strategy?!?!?!
You have to be either gianly myopic (figuratively speaking) or plain ignorant to fail to see the strategy behind Sun’s moves. Compared to the likes of IBM and HP the Sun’s strategy is very clear and focused. Lets review:
– IBM strategy: What OS flavor are we running today and what is the future? Is it zOS, AIX, OS/400, Windows, Linux on x86 or may be Linux on Power? What CPU are we going to pitch to be the future? Is it Xeon, Itanium, Opteron, or may be Power? IBM will try to sell you whatever is hot this minute with no regard for the long term strategy because the more complicated your environment is the money IBM can make from services doing the “integration” work. Bottom line is IBM has no strategy per se.
– HP’s strategy: Which dead-end OS are we peddling to the customer today? Is it Tru64, HP-UX, NonStop, OpenVMS, Windows, NetWare, or may be Linux and which Linux flavour is it going to be — RedHat or SuSE? What CPU strategy are we going to EOL today? Is going to be Alpha, PA-RISC, MIPS (NonStop), or may be Itanium and what are we going to use instead — Xeon or Opteron??? Truth is HP has got a giant mess on its hands and it is going to take years to sort this out, most likely at the cost of HP’s market share, which most likely will go to Sun and IBM.
– Sun strategy by comparison is clear and to the point: Use Solaris for all platforms (Sparc, x86), but if you really want Linux, we’ll sell it you. Use Sparc as hardware platform as it is proven, scales extremely well, and fairly cheap; but if you’re very price sensitive and want good price/performance on the low end, we’ll sell you Opteron. And oh yeah, just to sweeten the deal we can throw in gobs of free middleware IBM will charge you and arm and a leg for. Looks like a pretty good strategy to me, at least by comparison to IBM and HP.
All major vendors exept IBM are changing their cpu architectures at the moment – except IBM, because basically nothing can touch the Power5.
Sun : please run Solaris, look we will even let you run it on Opteron.
IBM : we’ve got these big, dependable (and expensive) machines here. Buy ’em from us and run what the hell you want on it. Linux, AIX, OS/400 we don’t care – run them all at once on the same server!
I personally prefer the IBM reasoning YMMV
I don’t get it. Why do people make it seem like Sun’s strategy is so difficult. Sun’s strategy is very simple. It is to sell you what you want. If you want an x86 system, they will sell you that system. If you want to run Linux on it, they will allow you to do it. If you want to run Windows on it, they will let you do it. If you want to use Solaris, they will let you do that to. If you want a behemoth of a system, they can sell you that too. They have their own products and solutions, but if you don’t want to run their product, they aren’t going to force you to use them. It is about freedom of choice. They will let you run your enterprise how you want to and will provide you with the capability of using Sun products if you wish.
>> Sun has figured out a way to make x86 hardware cheaper than anyone else
with 100% certainty I can say this statement is entirely false.
> IBM : we’ve got these big, dependable (and expensive) machines here. Buy ’em from us and run what the hell you want on it. Linux, AIX, OS/400 we don’t care – run them all at once on the same server!
I agree with what you’ve just said, only what you’re talking about is IBM’s midrange strategy. If you put together IBM’s low-end, midrange, and high-end strategies, you still get the same mess I was talking about above. I would love to see Power on the low end, but unfortunately it is not going to happen — the chip is too damn expensive.
>> Sun has figured out a way to make x86 hardware cheaper than anyone else
> with 100% certainty I can say this statement is entirely false.
It is not just making the hardware cheaper that allows Sun to beat Dell on price, but the fact that Sun owns the OS and middleware. Sun does not pay royalties on each box to M$ or RedHat/SuSE like Dell, IBM, and HP and can therefore easily make up the difference even if Sun’s hardware is a little more expensive from cost perspective. And again Sun can always sweeten the deals with discounts on middleware — something Dell or HP can not do. Sun x86 hardware is already cheaper than Dell, just check the prices.
” Use Sparc as hardware platform as it is proven, scales extremely well, and fairly cheap”
Cheap? Compared to what??
I think Sun will fail against linux but it is good have another heavy unix-based operating system on x86 market. Linux will gain indirectly with software-houses making more commercial applications for POSIX and X (for graphics) based operating systems.
I don’t see how Sun spent 500 million dollars on Solaris 10.
> ” Use Sparc as hardware platform as it is proven, scales extremely well, and fairly cheap”
> Cheap? Compared to what??
Yes UltraSparc has become a fairly cheap platform, at 4+ CPU’s Sun UltraSparc systems are actually cheaper than similarly configured Xeon systems and I’m not even talking about Itanium and Power, which are at least twice as expensive as Sun.
“with 100% certainty I can say this statement is entirely false.”
Since you are so certain, please honor us with your evidence as to why you are so certain.
You cannot cite Sun’s historical pricing strategies. They all changed this year.
“Cheap? Compared to what??”
POWER. Itanium. Perhaps even Xeon at the low end.
I remember when Sun’s v440 came out, people were amazed to see a SPARC server right in line with similar x86 servers. Sun also still sells the cheapest RISC UNIX server on the market (their $1000 UltraSPARC IIi boxes–nothing special but cheap for RISC). They also auction off new hardware at eBay, so you can even name your price. I seem to remember that McNealy said he really likes the auction model for sales, but that the market isn’t comfortable enough with it.
“I don’t see how Sun spent 500 million dollars on Solaris 10.”
Developing software is expensive. Think about engineers costing, say, $200K/year, including benefits, office space, computer equipment, etc. Multiply that by a thousand engineers, over several years. It gets expensive real quick.
im not so sure about that. they used to make the bulk of their money in their SPARC shipments. Their sparc shipments are down and their x86 shipments are up. They posted a tiny profit.
The x86 server market is extremely saturated. I sell x86 servers cheaply even. If they do become very strong in the x86 market what happens to SPARC customers? Rely on Fujitu?
What about sun’s increasing focus on software due to jonathan schwartz? What about his idea that everything in the world from toasters to servers should all be subscription based?
I think sun’s real problem is their marketing. I think they should keep their server line up as is and not increasingly focus on the x86 market. It seems they just advertise x86 instead of sparc.
So I’m skeptical. Sure I’m probably wrong and I have other beliefs of what MAY hapen but this is one of them.
I do ultimately think sun will win, or a company who buys out sun will win. In fact, schwartz even said they are considering any option right now. I am not sure what that means, but i think they will try to make a huge push somewhere or do something big but they don’t know what it may be.
“Any attempts to duplicate the Linux development model won’t work for Sun. Sun is releasing Solaris under an open source license, but it is mostly a marketing ploy to gain mind-share, and win Microsoft’s blessing. ”
They are assuming here. Releasing solaris as open source is benifit to microsoft? THIS IS NEWS TO ME. I disagree. They love linux (the execs personally) and think a huge open source communty behind solaris and developing it can cut sun’s development costs of Solaris and create a massive team of sun-loving coders that benifit sun. I don’t know what they mean as “development model”. That can be interpretated many ways. It seems like
they are just saying ..’ hey this is how it is ok, im right only’ I don’t like that they need to be more clear.
“Microsoft will be very supportive of Sun’s attempts to push Solaris as the open source operating system of choice. In the long-term a popular and technically superior Solaris will divert development effort from Linux ”
This is crap. Solaris competes against Windows and you know that windows is loosing marketshare to open source operating systems like FreeBSD and Linux. Did FreeBSD take away linux users? NO It got them from windows and new computer users!
“Sun’s overall strategy suggests that UltraSparc has no future. Sun has poured large sums into UltraSparc development in the past and is being forced to do so even now, but UltraSparc has been clinically dead for a long time now”
Crap. They see a huge problem in development costs and want to split development costs with fujitsu so they can keep the design alive. I mean look.
What is the most popular 64-bit CPU?? What ships more?? SPARC!! Getting rid of SPARC would be extremely retarded. They will just learn to manage
their SPARC business better. And let me remind you that there are more than one companies making sparc cpus and it’s an open design, more so than IBM’s POWER.
Have these people not relized that sun is trying to copy off of IBM and their compitetors? IBM is the most powerful server company and it’s growing. They are trying to copy off of IBM in the server world and have competing lines like IBM does. IBM has POWER, x86, etc. and sun will have the equvilent and will use it’s partnership with Fujitsu to actually develope SPARC better and compete with POWER (We all now power is much better right but that may change)
“will get swamped by hardware and software development costs and quickly go out of business. ”
lol.. sun spends so much money on software and then gives it away to the open source community. I do not think they care about wasting $400 million dollars on giving away solaris for free. They will cut a lot of Solaris’ best features and include it in the commercial version so they can actually sell it and incorporate the goodies from OpenSolaris as they are developed. They are also going from one server at a time, they aren’t releasing dosens of servers at a time.. as one server sells they release another one and so on. they are running it the smart way.. keeping development costs in line with volume.
Go talk to sun’s execs!
This article just tells this strategy these people firmly believe in without actually knowing or talking to sun’s execs personally. This is just an opinion article on some website. I do NOT take it seriously and if this article is 100% truthful i would say sun will die a painful death.
the sparc benchmarks may not be as good as IBM and Itanium but they comepte well with others. They sell them cheaper than ibm machines.
Lower TCO (Total cost of ownership) than other servers.
I really think the SPARC line would be extremely successful if they just got off their @$$ and got a better marketing team and actually used pretty graphics and marketed in the right places!
“but the fact that Sun owns the OS and middleware. Sun does not pay royalties on each box to M$ or RedHat/SuSE like Dell”
Actually, sun includes redhat and suse as an option when buying a server. This is exactly what IBM, Dell, etc. do
I personally think that people would be more likely to use other open source alternatives for their x86 servers than their SPARC servers (since they have a tight lock on those)
Sun isn’t abandoning Sparc. They are spending money overhauling it to the next generation.
SPARC development is now being done collaboratively with Fujitsu. With the exception of Niagra, the new generation of SPARC processors will use Fujitsu’s SPARC64 core. Fujitsu already has PRIMEPOWER servers clocked at 2+ GHz.
So, with that said and looking at Sun Microsystems as of late, their execs can’t seem to figure what they’re doing.
Javathan Schwartz may be clueless when it comes to PR, but he was behind Project Mad Hatter and Project Orion. Mad Hatter became the Java Desktop System, the single most successful enterprise Linux distribution (China dropped Red Flag Linux and standardized upon JDS, in addition to England) and Project Orion became the Java Enterprise System. Both have been huge moneymakers for Sun, and have reversed their previously gloomy financial outlook. Yes, the man who claimed that Sun’s Linux strategy is that “they don’t have one” is the man behind the single most successful enterprise Linux distribution. Bad at PR, good at business?
The bottom line, Sun is making the right moves right now after years of post-dotcom bubble fumbling. To say they don’t have a strategy is simply naive, their stock value certainly says otherwise.
>> Sun has figured out a way to make x86 hardware cheaper than anyone else
with 100% certainty I can say this statement is entirely false.
Sun is the cheapest tier 1 Opteron vendor. Considering all of our purchasing contracts are with tier 1 vendors, it’s an immense amount of work to get purchases through 3rd party vendors through our purchasing department. YMMV where you work, but in general it’s rather important in the enterprise.
Sun is supporting Solaris, Windows, or Linux on their Opteron servers. I think this move towards diversity comes as HP is eclipsing their Unix server sales with the choice of Alpha, PA-RISC, and Itanium platforms running your choice of Tru64, HP-UX, or Linux. Sun sees the potential for the Opteron to eclipse at least the latter two architectures (and eventually Alpha as HP is determined to see it die at the hands of Itanium)
Sun currently leads the tier 1 players in Opteron sales as well… I think as Sun diversifies their Opteron offerings we’ll begin to see the 3rd party players marginalized and eventually we’ll be left with an Opteron battle between Sun, HP, and IBM (unless Dell wishes to enter the fray)
Linux is failing badly. Other than RH, the distros are not compatible. Too many pieces are broke and inconsistent from one to the other — and RH, Solaris 10 makes RH way too expensive.
If you actually put a Linux desktop into production use, the failure becomes obvious. Who has the time to continually try and keep up to date on a lot of machines. The concept of update often is a deal killer for business. My desktop distro of choice for clients is SUSE — yes it has the same problems. I am slowly switching to JDS — it has less of the problems.
Sun and IBM understand a business needs stability and long term use. Solaris will bring that to open source. Give me drivers that have been stressed and tested — as opposed to drivers that sort of work some of the time. Hardware is so cheap I can afford to buy Solaris supported X86 hardware that works. Anyone ever bother to check the packet error rate on a $10 enet card running the supplied driver.
Who cares if I can donwload the latest release of who-so-waz-it software and watch it crash. Find that OOS package which installs with the ease of windows XP. Does anyone think that joe-six-pack-computer-user can open and install a tar file?
Linux’s success is becoming it’s failure. Linux makes the 80s *NIX wars look like child’s play. Crash, bang, splat — big time.
SPARC development is now being done collaboratively with Fujitsu. With the exception of Niagra, the new generation of SPARC processors will use Fujitsu’s SPARC64 core. Fujitsu already has PRIMEPOWER servers clocked at 2+ GHz.
This statement isn’t true either. Sun and Fujitsu are collaborating to build servers, not microprocessors.
http://www.theregister.co.uk/2004/06/11/sun_server_fujitsu/
If you read this article it states that both Sun and Fujitsu microprocessors will be used in forthcoming server designs that will be shared by the two companies. Both companies are continuing to develop SPARC processors, and will do so for the forseeable future.
I find this article to be an interesting piece of work; however, I think the author has speculated too much in situations where he shouldn’t. The article contains some conspiratorial and factually incorrect claims, and some that I just disagree with as a matter of opinion.
In particular:
Some of Sun’s x86 gains will surely come at the expense of its high margin Sparc business so the company has to compensate for that loss.
This is something that I hear all the time, and it doesn’t reflect the fundamental reality about why Sun got serious about the x86 server business. The suggestion that selling x86 solutions comes at the expense of the SPARC side of the business is specious. Selling x86 solutions actually allows Sun to keep business that they would otherwise lose to companies that are selling x86 hardware. One of the reasons Sun had been doing so poorly once IT companies’ spending was constricted, was precisely because companies had less money to spend and wanted to buy x86 hardware. By not offering x86 solutions, their customers (or prospective customers) had to go elsewhere and obtain other products. Sun doesn’t lose business by offering x86 hardware, rather they keep business that they otherwise would might have lost.
Also, I would call SPARC comparatively high-margin with respect to x86. I don’t think anyone can afford to sell microprocessors at really high margins.
A price war will only hurt Intel’s profitability as AMD isn’t making too much from server processors.
AMD has been making lots of money from Opteron. I don’t understand the basis for this assertion.
[T]he company will likely need to support an Itanium port as well.
Sun isn’t going to do this. They attempted to port Solaris to Itanium many years ago; however, Intel hampered the process and screwed Sun over at every turn. The port was never completed, and Sun has no interest in reviving it. There also is no plausible business justification for doing so. Sun already has a cheap x86 compatible 64-bit platform.
Dell spent $464 million on R&D in all of 2003, and Dell is much bigger in terms of revenues than Sun.
Dell doesn’t need to do any R&D, which is why their budget for it is *so* small. Comparing Sun to Dell in this respect belies a rather serious misunderstanding about the differences between the two companies. Just as another point of comparison, Microsoft spent almost $7 billion on research in 2004:
http://www.fastcompany.com/magazine/89/microsoft.html
If anything, the question to ask wouldn’t be why is Sun spending so much on R&D, but really, why aren’t they spending more?
Sun is releasing Solaris under an open source license, but it is mostly a marketing ploy to gain mind-share
This is a disingenuous statement that is unduly cynical. I’ve met the engineers responsible for getting OpenSolaris off the ground, and they really are interested in building a community around OpenSolaris. Whether this will actually happen or not has been debated here ad nauseam. However, I think it’s unfair to say that this is *only* a marketing ploy.
If Sun manages to sell millions of servers, Solaris development costs will get dispersed over the large number of units shipped and become irrelevant. Also, Solaris will displace Linux as the open source operating system of choice, and this will allow Sun to steal IBM and HP’s Unix customers.
Volumes are critical; however, the suggestion that massive x86 volumes will obviate Linux is laughable. Linux is here to stay, and people should avoid getting stuck in the simplistic zero-sum OS game. The success of Solaris x86 does not come at the expense of Linux. Also, Sun will ship their x86 servers with Linux or Solaris, so volumes sold doesn’t have a direct correlation to which OS is being installed. Further, Sun has been qualifying their x86 servers for Windows installations, so it’s equally plausible that a customer with a Windows license could purchase x86 servers without any OS and install Windows later.
However, lack of x86 support for AIX and HP-UX may be an issue for IBM and HP customers that Sun could exploit to their own advantage.
Microsoft will be very supportive of Sun’s attempts to push Solaris as the open source operating system of choice.
I seriously doubt this. Microsoft believes that Windows is the one true operating system. Microsoft also derives a lot of financial benefit from software development on their platform. I don’t think that they would push anything other than Windows. Microsoft and Sun are still competitors, I wouldn’t count on MS helping Sun out.