Sun Microsystems is spitting nails over a biting report from Meta Group, a market research and consulting firm, that suggested, among other things, that Sun’s customers should rethink doing business with the company in the long term, and that CEO Scott McNealy should be replaced.
Sun Microsystems Fights Back After Caustic Report
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
The report is right on the money. The situation with SUN and the Intel market is generally the same as that being faced by Apple and its dwindling niche market, except that, while Apple is actually trying hard to adapt into other markets, SUN has adopted a “too little too late” approach. Sun and the Apple PC market think that people will keep paying ludicrous sums for hardware just to gain marginal advantages. Its a shame to see a formerly innovative company like sun going to the gutters. Why the hell is McNeally steal CEO? The man ran out of ideas a gazillion years ago.
I should know, I work with it on a daily basis =)
I would have canned McNealy years ago. Ever since his first reactions towards Linux.
Sun has technology and would have fit perfectly into that OSS model if only they had management that could see farther than their next paycheck.
of people who have obviously absolutely no experience with Sun environments, hardware, or Solaris commenting on stories relating to Sun.
Sun and the Apple PC market think that people will keep paying ludicrous sums for hardware just to gain marginal advantages.
The problem is in many cases there are no alternatives to Sun besides a less reputed vendor offering a similar solution. This is the case for mission-critical task that hinges around some centralized resource.
While there are Linux solutions being offered for these sorts of purposes (SGI’s Altix) they simply aren’t selling. Furthermore, Linux offers a fraction of the features of Solaris for minimizing downtime. On any higher end Sun system, you can remove virtually any component (besides the system mainboard, of course) while the system is active. This includes things like CPUs, memory modules, and peripheral cards. This requires a degree of hardware/software integration not easily attained with something as cross-platform as Linux. Diagnosis of faulty hardware is also much easier on Solaris. On any version of Solaris with the scrubber patch, it is possible to determine what hardware components are faulty from within Solaris itself.
These are just a few examples of why Linux isn’t ready for big iron systems which are home to mission-critical tasks. That’s not to badmouth the Altix, it looks like an excellent system, and it’s great for its target market: parallel computing for scientific tasks. IBM has begun to push Linux on their big iron, and we’ll have to see what modifications IBM will make to it to provide support for servers running mission-critical tasks. Unfortunately, this is all vapor at this time and can’t even be assessed.
Sun is a company with a wonderful long-term vision. They are attempting to develop resources to compete head-on with Microsoft in the corporate environment, leveraging whatever tools they can. They aim to provide a solution that eliminates a great many problems with the current Microsoft endorsed approach: thousands of power-hungry, hot, and noisy PCs running Windows connecting to relatively underpowered x86 servers. I’m talking, of course, of a Sunray environment, which would centralize all the computing resources of a company, decrease power consumption and environmental costs of computing, and allow for users to seamlessly move between computers.
This is a great idea, unfortunately the applications software simply isn’t there to make it happen. Sun has no idea how to solve this problem. They’ve been looking to Linux but, unfortuantely, Linux suffers from the same problems. They have no direction because of this… they’ve developed all of the technology necessary to start selling Sunray environments, but no one wants to buy because the drawbacks outweigh the benefits.
My understanding was that the thin client (sunray) model was currently enjoying a renaissance as the intel sever price was making PCs look expensive where the big iron Sun sparc price made PCs look cheap.
Sun’s recent “talking up” of solaris on intel would imply they see a long term, price compelled, shift to intel.
Please, if the only reason you don’t like SUN is because of their take on linux, don’t say McNealy needs to go. This is simply a business decision that he believes will pay off. Linux is the “new thing” right now. He is right on the money. Eventually, it won’t be, something else will come along and take the “new thing” area over. Remember, Hindsight is always 20/20. Its easy to take cheap shots at somebody after the fact.
“Mission critical.” How many times has that awful phrase sunk companies? I’m sorry, too many companies have been sunk during the dotcom days just by paying through the nose for a bogus Mission Critical box of stable junk.
In the real world, people prototype on inexpensive systems because they can’t justify the risk of throwing resources around on stuff that might not be needed. Then occasionally one of these systems is the bedrock of something important, and it stays on the cheaper platform, using redundancy and operator intervention to get stability.
One day, someone says, “This system holds itself together with string and duct tape. Rewriting this with a real design should take no time at all!” Years later, the porting project completes 3X over budget and with major reliability problems.
Are there Sun customers who know what they’re doing? Sure, most of them do. But people who know exactly what they want do not represent the kind of market that would make Sun a great investment. The main thing helping Sun out is they probably have a fundamentally sound business as long as they don’t want extreme growth rates. I’m sure their stock will go up.
I work for a big company that is a major IBM shop – true blue – through and through.
Except IT here wouldn’t consider running their webserver on anything else but a Sun box. Nothing else is as reliable. AIX *cough* *spit*, right
I noticed two interesting things here:
$8.4 billion compared with $9 billion a year ago
Based on that, I’m not sure what all the fuss is about. The other thing is this:
[Sun is] being squeezed by Dell Computer and others at the low end
Now correct me if I’m wrong, but Sun has only recently entered into the low end market. How can they be squeezed out of something they’ve just gotten into?
It seems to me obvious that Sun is going to lose revenue. They way they can get more competetive than they are now is by lowering their prices because low end hardware is getting better. Sun is a technology leader and will continue to be. With a complete line of lower end hardware that they’re developing, the only thing that will give them trouble in the future is marketing. Unfortunately, that means more than it should in today’s world.
I work with Sun alot, I work with alot of companies but Sun is a great group of guys to work with. But yes Sparc and Solaris is a niche market and I notice our Solaris customers are dwindling big time. I think Linux adoption is neccessary at this point, because Intel and Linux are making some serious inroads, and yes Bascule, I have customers that use Linux on the high end and do very well with it. Their new initiative N1 is very confusing as well and it just makes IT work harder because we now have to keep up with the Model quarters and its a mess right now. But the Meta group did a very fair review. If Sun doesnt like the response then why did they hire them? They shouldnt be getting mad they should fix it.
i’ve worked with ibm x series running red hat. what’s this about sun being the only high end server that you can swap everything out while the system is still running? i’ve done it on an ibm x series before. faulty rams, just removed and replaced them. needed extra harddrive space, just added. it even has lights inside to tell me which part is faulty.
ohh…lights…do they flash ? or are they red/green/yellow, yellow being…almost dead component? lol
yes. and it blinks blue and red during christmas.
I’m so frustrated at trying to install applications to Solaris 8 for Intel right now that I don’t care if they go broke tomorrow! Software on the companion disk has to be *compiled* first!?! Kiss my a$$, Sun.
That’s after the ass-pain of trying to install the OS in the first place. The solution was to give up trying to install on my primary computer and install on backup machine with only other OS being Win98. Still haven’t found graphical tools for hardware configuration, user-management, or package installation. Those have been standard on Linux distros for *years*.
Maybe Solaris 9 has “fixed” these things, but I’m not paying twenty bucks to find out.
Maybe I’ll have better luck now that I’ve downloaded the compiled freeware packages. Can’t unzip files from CD – they have to be copied to a certain location on hard drive? Required libraries aren’t included in packages and have to be installed seperately? Next week sounds like a doozy too…
-Bob
Seriously,
Would someone decide his mission critical system choice be based on whether it has a cute installer or not?
Didn’t think so neither…. the fat penguin is just a mediahype, not a ninja penguin!
Sun hasn’t delivered any working innovative technology to the market in years.
No one can even comprehend what “N1” even means. And it is not like Sun has enough clout in the datacenter to make everyone support their proprietary protocols.
What else does Sun have besides “N1” ? More versions of Java that offer mostly just bloat and bugs beyond a few minor improvements?
Obviously from the retards that have been interviewed on OSNews, Sun doesn’t have anybody in management who is competent. Or even honest.
The only chance Sun has is to purge their management ranks and get some intelligent people to run things. And then to pick some projects that deliver real value to the customer, not more vendor lock in and consulting dollars for Sun.
I agree. If McNealy why not also Schwartz?
If anything is left when the fire is out then it is Java.
Why not take Java and make something more out of it than just a backroom server. Softwares for home user, office and enterprise from A to Z. On Linux on Sparc. And slowly retire Solaris. As it looks like IBM eventually will (have to) retire AIX. Java not fast enough? If so why did Microsoft develop their equivalent CLR? SUN has an advantage that DEC did not have. The fate of a very similar company to learn from. Despite a superior CPU – the Alpha. Why did IBM survive? Maybe it was the change of leadership.
Ok the review might have been hard. But Sun has died. I have used various architectures and processors. And in the end it is an Intel Instruction Set World. It might be a sad thing that we only use one instruction set, but that is how it is.
The advantage is that you have critical mass. Eg you can get parts cheap, etc. Now about reliability, lets talk about that. You can either get a big expensive box or many little boxes. Gee, hmmm, what is Google doing? Oh yeah many little boxes. Many little boxes are less expensive and work just as well as one big huge box. The future is grid and clusters…. Sun just does not want to understand that…
Sun has been “dead” since the time Intel rolled out the Pentium, just like Apple. Oh, and IBM was supposed to have died in the mid/late 80s. I must say that I am quite impressed by all of these zombies.
Hey there is an economic downturn, deal with it. Even FeeCee vendors are feeling the heat.
For every job there is a tool, I am tired of all these PC-windoze/linux fanboys trying to claim that the one size x86/windows/linux fits every need. Sure Solaris is a nice player, and they never intended it to be otherwise. But Linux is also a niche player, so deal with it.
I really love when people make statements like these:
“The future is grid and clusters…. Sun just does not want to understand that…”
Hello, you know what is SUN’s corporate motto? THE NETWORK IS THE COMPUTER. If any company understood distributed computing it is SUN. SUN is one of the few vendors offering packaged grid and clustering software and hardware solutions. Go and download SUN grid, or play with the built-in clustering in Solaris. But please do not stop things such as reality and facts stay in the way…. no siree bob. Ignorance is bliss, I guess.
That is why I keep suggesting that SUN should buy Corel and port their applications over to Solaris. At the end of the day, they promote centralised processing and thin client technology yet there are no client applications out there that run on their hardware. OpenOffice is alright, however, Wordperfect, Paradox, Quatro Pro and Presentations would be the icing on the cake for any organisation wanting an alternative to the Wintel world.
As for the porting, just have a look at http://www.mainsoft.com and how easy it is to port win32 applications to *NIX without years of re-writing.
anon: “Mission critical.” How many times has that awful phrase sunk companies? I’m sorry, too many companies have been sunk during the dotcom days just by paying through the nose for a bogus Mission Critical box of stable junk.
A “bogus Mission Critical box of stable junk”. So what exactly are you incinuating, that all computer components are junk, or that an UltraSPARC III is a piece of junk when compared to your latest and greatest Athlon or P4 processor?
follerec: i’ve worked with ibm x series running red hat. what’s this about sun being the only high end server that you can swap everything out while the system is still running?
I specifically mentioned that IBM would (soon) support it but that the necessary OS modifications have not been completed at this time.
i’ve done it on an ibm x series before. faulty rams, just removed and replaced them.
Well, I don’t know what model of xSeries you own, or what you’re saying specifically. What particular copy of RedHat were you running? And do you claim that you had an active system, then you removed RAM and replaced it without rebooting the active system?
Michael: Sun hasn’t delivered any working innovative technology to the market in years.
Umm, how about Sunrays and the Sunray Server? They may be just glorified thin clients, but they’ve been glorified to the point that they’re actually usable for day-to-day tasks.
What else does Sun have besides “N1” ? More versions of Java that offer mostly just bloat and bugs beyond a few minor improvements?
The UltraSPARC IV, V, and Niagra, all with extensive support for SMT. Support for this will tightly integrate with virtual instances. Any applications with a set of threads manipulating a similar set of data (i.e. databases) will greatly benefit from this arcitecture.
Christian Gross: Now about reliability, lets talk about that. You can either get a big expensive box or many little boxes. Gee, hmmm, what is Google doing?
Google started with virtually no financing, and consequently they have a power-hungry legacy architecture of distributed nodes, as they simply built on top of where they started from. Google is probably the poster boy of a company that can benefit from an SMT-enabled architecture that supports cache coherency. It would not surprise me if a cabinet of Sun Niagra processors could manage the entire processing load of google.
Google developed software in-house that would run on such a distributed environment. For the majority of the world this isn’t possible, and if database scalability or nightly compile times need to be brought down the only solution is to buy a bigger machine.
Oh yeah many little boxes. Many little boxes are less expensive
Not in terms of TCO. Clusters are power consumption and environmental nightmares. Your userbase may be happy, but you’re going to anger your building manager and the HVAC guy. And just wait until the air conditioner goes out and every system in the server room overheats.
Consolidation makes for a significantly more maintainable environment.
and work just as well as one big huge box.
As I’ve detailed… they only work as well as one big box provided the task is one that can easily be distributed. This isn’t the case for things like DBMs.
The future is grid and clusters…. Sun just does not want to understand that…
And let ignorance shine on through! Oh my, Sun behind the times on grid computing? Well, isn’t that quite an allegation…
Son, I suggest you read up on SunPlex and N1. You picked as a weakness one of Sun’s greatest strengths.
“The problem is in many cases there are no alternatives to Sun besides a less reputed vendor offering a similar solution. This is the case for mission-critical task that hinges around some centralized resource. ”
If I look at SUN’s shares (falled 90% in the last couple of years, btw. this is reason enough to fire McNealy) and sales there seem to be a lot of alternatives which get used. If there would be no alternatives then many customers would have to use SUN’s expensive products and SUN would be in much better standing. I don’t see any cool inovation from SUN which should turn the wheel.
Based on personal experiende with two Sun Fire v880 servers at work:
Sun has poor performance.
Sun machines are EXTREMELLY expensive.
Sun spare parts are SUPER EXTREMELLY expensive.
😮
I always hear that argument. “Look at Google, they use a bazzillion white box PCs made from scrap!”
The dark side is that most applications are not built in a way that works well with the Google model. Google does a pretty specific thing compared to most corporate information systems. Google doesn’t have to provide instantaneous updated information to all of its users. Google can scrape a web page and not publish search results for a week, and noboby would know or care. Most inventory systems don’t have that luxury. Google can present inconsistent results to multiple people asking the same question and not get screaming phone calls from customers or the shipping department.
Modern corporate systsms do a vast array of things and must compromise their system design to accomodate both current needs and the whims of management tomorrow.
Also, inevitably, all of the traffic ends up getting focused on a single machine, the back end database. The more clients you have out front pre-processing information, the more responsive the final database needs to be, otherwise all of that cheap horsepower comes to a screeching halt.
I continue to put a lot of thought in how to “scale wide” versus “scale up”. When you have two 4 CPU machines running server software, and the prescribed method for scaling is to cluster the machines, AND the software licenses to do that run ~$100K, you think of all of the hardware you can buy for $100K and try and figure out how to make it work.
But if your application doesn’t consider this model up front, it’s really difficult to actually port it later. Lots of network latency issues, replication and cache cosolidation. More race conditions than NASCAR. I don’t care how fast your Porsche is when its in gridlocked one lane traffic with a plethora of stop lights waiting for it.
In the end, it inevitably boils down to the database though. The DB ends ups doing most of the work in the long run, and everything else just feeds it. The more feeds it has, the bigger the box needs to be.
Google is a fluke, so is Pixar.
Of course there is frustration that the whole world isn’t using the particular hardware setup that YOU use… These types of threads come up all over the web…”Company X is sinking, because they only have 3% marketshare”…then in another thread the same person rants “MS stinks because the have 95% marketshare.” Hasn’t anyone learned that computers are tools and we use the tools that get the job done AS WELL AS the ones that we’re familiar with? Come on, it’s natural.
But, with Java, Sun has a real chance to create a software development revolution. They just need to actually focus on it and improve it’s indescrepencies (make swing faster, decrease jvm bloat, and the many others that many of you can add…).
If Sun put some peoplepower behind Java, we could really see it take off on the client side as it has done on the server side. I think they are on their way, but I wish they would speed up their pace a little.
PD
Anonymous: If I look at SUN’s shares (falled 90% in the last couple of years, btw. this is reason enough to fire McNealy) and sales there seem to be a lot of alternatives which get used.
And despite all the death knells being sounded by people like you, Sun turned a $4 million net profit last quarter?
shark Based on personal experiende with two Sun Fire v880 servers at work:
And what “experiende” is this? Are you the system administrator? My guess would be no. If you are a consumer of services these systems provide, what leads you to believe the servers are the source of the problem?
Sun has poor performance.
Care to define what this means? Are you running some sort of CPU bound task and then complaining about its performance?
Sun machines are EXTREMELLY expensive.
Well, I’m not going to argue that Sun hardware isn’t “EXTREMELLY” expensive, especially if you’re purchasing it directly from Sun and aren’t negotiating the price. As I work for a university, we receive a discount on all purchases. Furthermore, there are several resellers of Sun hardware which can get you a discounted price.
My original point stands: there are tasks for which the expense is justified, and if none of these tasks fit into your business model, there is no reason to purchase Sun hardware. However, it’s pointless to argue with people who do have legitimate reasons for using it, stemming from your obviously tangental personal experience.
Sun spare parts are SUPER EXTREMELLY expensive.
So don’t purchase them through Sun. There are a number of companies that stock Sun spare parts, like Western Scientific, http://www.wsm.com
Or you could… buy a service contract. That’s what they’re for.