I couldn’t resist this post from Jaime Cardoso, ranting about yet another questionable Gartner claim.
“Some people say that Gartner should start blogging and join the conversation. About this, I say Thank God for small favors since I believe engaging a conversation with a drunk in the nearest bar would be far more profitable.”
What triggered Jaime’s ire?
Donald Feinberg, vice-president of Gartner, just claimed: “Linux is coming, Unix is dead.” Apparently this is an “absolute”.
So Unix is dead, eh? More binary reductionism. Ever hear about the Long Tail? Ever looked at the history of IT industry, and the fact platforms hang around for an awfully long time, establishing valuable ecosystems in their own right? Ever considered that making claims like this might affect your credibility?
Its clearly factually incorrect to say that Unix is dead in 2005, and its highly likely Unix will win some new production workloads in 2010, particularly in scale up, OLTP scenarios. The IBM mainframe after all is still winning new workloads, and it has been “dead” for a long time…
“Linux and Windows will be the only two operating systems left in five years.”
Its surely dangerous for analysts to fall into the trap of binary thinking. It doesn’t make for nuanced analysis, just FUD heavy headlines. My biggest question though is just how does such a statement help the end user community? If you’re a Unix exponent you just got shafted by the biggest analyst firm in the business. By all means recommend that enterprises should apply more resources to Linux and Windows, but why play the dead card?
As Stephen says:
On one level, the problem with such arguments is obvious: they simply do not reflect reality. Enterprises, for example, do not typically use a single platform – Java or .NET – for their development needs. They’ll use both, leveraging the different strengths – not to mention a host of other languages and platforms like Perl, PHP or Python. Distilling the question down to Java v .NET has the benefit of focus at the expense of utility.
Even worse, however, is the fact that binary arguments tend to obscure the fact that in many real world implementations, so-called oppositional technologies will compliment rather than annihilate each other. At the very least, they provide each other with the competition that drives innovation.
Competition driving innovation. Like Linux communities attempting to emulate DTrace in Solaris 10, say.
IDC is tracking OpenSolaris. Is Gartner? We are – here is a Q&A. In order to understand a platform’s potential longevity its surely important to consider things like the strength of the community, and the platform currency of customers. Brandon Werner is having so much fun with OpenSolaris that he has renewed his vows to C++. Being in love with one “dead” platform is unfortunate, but to love two “dead” platforms surely smacks of carelessness…
So is Unix really dead?
Karl Freund, IBM vice president, pSeries offering management, do you think Unix is dead? Or will you ship AIX 5.5 in 2010? Catherine? Maybe I should ask the AIX expert blog?
What about Sun? Bryan? Jonathan?
HP? Who is “the face of HP-UX”, anyway? [Could be a problem for HP. We know who Martin Fink is but…]
Ben?
I am little worried that RedMonk’s industry analysis is off the mark, though, given we think IBM and Microsoft are going to play industry shaping roles for years to come. Jaime has his own dead platform 8-ball:
Wait, wait, don’t go away right now, I’m having a vision, I’m having a vision, … I see, … I see, … IBM is DEAD and, so is Microsoft!
Jaime Cardoso says:
September 27, 2005 at 3:00 pm
The thing that infuriates me the most is why we all keep on mentioning Gartner has a source of analyses. Does anyone even remember a Gartner opinion being usefull? Yes, in terms of sales, they are #1 in the analyst market but, since when is market share a good pointer to quality? Does the Finantial Times sell more than the National Enquirer? (Sorry, those are the only 2 newspapers I know outside Portugal).
The same way, if you get all your news from National Enquirer, Alien Abduction is the #1 cause of dead in the US, If you make analyses based on googlefight you get, … a Gartner Analyses.
christopher baus says:
September 27, 2005 at 5:21 pm
I agree with your point that too often tech discussions are over simplified with abosultes. I do this myself, which is probably the result of years of binary thinking.
But what do you mean by Unix? AIX isn’t Unix. HP-UX isn’t Unix. Solaris isn’t Unix. The only “real” Unix left is probably SCO UnixWare, and for all practical purposes that is dead.
Everything that is left (BSD’s, OS X, Linux, Solaris, AIX, etc.) are some sort of Unix-like, POSIX implementing, system.
What I think you mean by Unix are commercial implementations of Unix-like OS’s. The term Unix isn’t dead, but maybe it should be.
James Governor says:
September 27, 2005 at 5:30 pm
wow christopher that’s what i call opening the question up. but you and I both know, I suspect, that is not what Gartner meant.
yes i do mean commercial implementations… which actually are Unix, and licensed accordingly, although they do have significant extensions.
If we’re going to say AIX, HP-UX and Solaris aren’t Unix flavours then the conversation becomes very difficult indeed.
Jaime Cardoso says:
September 27, 2005 at 6:18 pm
Definitions aside, is it reasonable to think Gartner was talking only about SCO? I don’t think so. But, a definition that offers very little room to interpretation is the definition of Dead.
Under ANY definition I know, the Mainframe isn’t Dead, neither is Smalltalk, Cobol, Unisys, Xerox. Not even Itanium (or HP itself) is dead.
SCO is dead? So, how come I still see it in some customers and I see ISVs supporting it?
Dead is Dead. Solaris is far from Dead, AIX is far from Dead and, even **IF** HP drops HP-UX and VMS, it will take a lot of time until they actually die.
BTW, wasn’t it a Gartner consultant that was “invited” to do the opening in the Itanium2 announcement.
Bryan Cantrill says:
September 27, 2005 at 9:16 pm
Feinberg’s comments are (of course) pure jack-assery — and they reflect a complete misunderstanding of the persistence of software. Even if we weren’t actively developing Solaris (which we are) or actively taking business away from competitors (which — much to Gartner’s surprise I would guess — we are as well), Solaris would not be “dead” by any stretch of the imagination. Software — especially larely correct, widespread software — exists in perpetuity. This is true of commercial software — and it is tautologically true for open source source software. But it apparently takes real analysis — and not the simple reguritation of Slashdot headlines — to see this truism. James, you must therefore find Gartner’s belly-flop to be quite reassuring, if not outright entertaining. It brings to mind something that I have said often in the halls in Menlo Park: “we’d be screwed were it not for our competitors…”
Catherine Helzerman says:
September 28, 2005 at 3:48 pm
Ah, the old UNIX is dead comment. I’m a Linux user and I’ll be the first one to tell you it’s a great OS. Having said that, there are several reasons it will not replace UNIX any time soon. I’ll quickly mention one. Linux is open and community developed. This model has provided customers with a free or low cost OS, quick bug fixes, and support for pretty much any platform out there.
IBM and other companies contribute code to Linux. Some is accepted, some is not. A feature that makes AIX a stellar OS on Power, might not benefit, say, a customer running the Linux kernel on his laptop, so the feature may rightly not be accepted. Because of their proprietary nature (that’s right, I said the P word), operating systems such as AIX and Solaris can be developed with their individual customers in mind. Sun doesn’t have to check in with a community to see if D-trace will run smashingly on everything…it just needs to do the job for Solaris customers. Ditto for AIX, which has many advanced features that make it a quick beast on Power but might not be practical elsewhere.
Will Linux *ever* replace Unix? I have no clue, but it won’t be in the next five years.
James Governor says:
September 28, 2005 at 4:18 pm
insightful, open and honest catherine. thanks.
I am obviously in complete agreement with Bryan about jackassery… and how its great when competitors make your life easier.
Ludovic says:
September 28, 2005 at 6:06 pm
You probably mean “complement” rather than compliment?
http://www.onelook.com/?loc=bm2&w=complement
Ludovic says:
September 28, 2005 at 6:08 pm
Great post James, I agree on the aspect of binary thinking. Many analysts are stuck in their silos and seek headlines to get coverage…
Jonathan Steel says:
September 29, 2005 at 10:56 am
James – you and I both know that Gartner works on the same principle as Forrester’s founder George Colony (as he was once quoted) – “it’s not about being right, it’s about getting the headline”.
Nuff said.
John Clingan says:
September 29, 2005 at 5:21 pm
James, I really think that by making that comment, Gartner is trying to get some attention. The larger question is regarding accountability. In 2010 will Gartner be accountable for what they said in 2005? What form will that accountability take? Revenue? I am with firm belief that we need a rating system for industry analysts based on accuracy. Perhaps there is one and I don’t know of it. Rating systems exist for financial analysts. We need a similar system for industry analysts. Not sure exactly what the metrics are, but with a system in place there is a .9 probability that junk comments such as Feinbergs would be addressed before they are even made 🙂
Gary Edwards says:
September 30, 2005 at 12:35 am
Correct me if i’m wrong, but doesn’t Gartner worship at the SOA alter?? Are they having a Janus moment?
The promise of SOA is fluid interoperability in a heterogeneous environment. With SOA there is no need to be bound by the platform specific demands of the past that so crippled client/server architectures.
The key to the future isn’t going to be found in the cloud of traditional Gartnerian guestimate methods of looking at marketplace uptake of the moment, and then proclaiming grandly which platform will live and which will die. The key to the future is going to turn on which platforms chose transparent interoperability methods and implementations as the foundation of the value base of what they provide consumers. The emergence of SOA ESB dramatically changes the decision making process for evaluating platforms.
So if Gartner believes in SOA, they also must believe in the fluid interoperability of Open Internet Technologies that underly all SOA. Things like support for Open XML technologies and Open Standards. And support for Open Interfaces and Open Messaging and Communications protocols. Support for Open XML technologies like OpenDocument is critical for consumers trying to construct complete SOA based next generation infrastructures.
If Gartner believes that all roads lead to SOA, how is it that they also conclude the Windows platform will live? The Windows XP integrated stack model is adverse to everything that makes SOA so valuable, so important.
Once again we see that legacy marketshare, platform control, and the influence purchasing power of a wealthy monopolist fog Gartner’s vision of the future. It is the Open Internet that is pulling the SOA ESB wagon. We know that Linux is an Open Internet ready platform. We know that Microsoft is determined to control with an iron grip how the integrated WinXP stack interoperates with anything outside of the Redmond realm. That includes the Open Internet. Does anyone really think that OpenSolaris isn’t going to be a better Open Internet SOA citizen than Windows XP Vista?
Even Gartner realizes the Challenge Chairman Bill faces trying to force march the entire 400 million Windows desktop user base to the integrated Windows XP stack model unleashed in the 2003 series of applications, OS, developer tools, server suites, and .NET framework. The truth is that most of the great herd of Windows users have been orphaned. If they want next generation collaborative computing functionality, the great herd has two choices. They can upgrade their hardware, software, and server systems to meet the costly platform demands of the integrated XP stack. That migration seriously compromises any hopes of adopting a fluid SOA loosely coupled integration of legacy, current, and future information systems. If you choose the XP stack, it’s best to choose it all, pay the price, rip out and replace everything with MSware, and kiss your interoperability with the Open Internet and the emerging digital civilization that thrives there good-bye.
Or, they can go about the business of downloading an SOA Open XML friendly core of a next generation collaborative productivity environment based on OpenOffice.org and Mozilla components. Maybe they will like a StarOffice or WorkPlace implementation. Soon enough KDE KOffice is going to arrive too.
The point is that the great herd of Windows users can have access to next generation collaborative computing without paying the disruptive and high cost of being force marched into the binding entanglements of the tightly bolted, highly integrated, iron coupled XP stack.
So which face of Gartner will the marketplace listen to? The SO2A face, or the Windows face? You can’t have both because sooner or later, one will cancel out the other. IMHO, the only platforms left standing will be those that embrace the Open Internet, Open Standards, and Open XML technologies. And one of those platforms is going to challenge Microsoft for the collaborative computing future of the 400 million Windows users Chairman Bill has left behind. And left up for grabs.
There is something about this proclamation from Gartner that only Linux and Windows survive that reminds me of a scene from the movie, Jurassic Park. After a long winded day of arm waving and grand proclamations about the glorious financial future of the Park, there comes that scene when the lights go out, the rain is pouring down, and the darkness of the unknown reigns. There is a giant thud. So powerful that tables and chairs shake. Glasses of water tremble. Blood in the veins of the arm waving experts suddenly chills. The rain water that has puddled up in the well of a giant clawprint shivers in fear as another thud sounds. The unknown darkness is about to explode into a future no one dared dream possible. T-Rex has arrived. And her name is Google.
How could anyone ever think that the next generation of collaborative computing is going to be anything like the past? Old rules don’t apply. And neither does old thinking.
~ge~
Mike Dolan says:
September 30, 2005 at 3:25 am
Does anything in technology ever “die”? Last I heard there are still OS/2 user groups…
The only thing that seems to be “dying” today is monolithic code bases and vendor lock-in (much of which has been in response to Linux/OSS changing attitudes towards technology).
Sam Wintergreen says:
September 30, 2005 at 6:14 pm
Eight track tapes and VW bugs are not dead either, but in a sense, they are. Feinberg is speaking with a bit of lattitude I’m sure, but I take his point to mean that within 5 years (or 15 depending on the quote you read), new application development investments will not be focused on the legacy Unix OSs. Linux will be the “no brainer” decision of the day. For those making new choices now – this “insight” is useful. I don’t assume Gartner is always right, but appreciate their point of view. -sam
Jaime Cardoso says:
September 30, 2005 at 6:54 pm
Gary. I ear you but, the thing is that Gartner can bet on SOA and on a single OS in the Market (or two) because they don’t even understand the issues involved.
If you (Gartner) take the technology out of the technological decisions (contradiction is intended to acentuate the stupidity), all you have left is the newspapers. Gartner get’s away with things like this because people that give them money don’t understant technology anyway and can’t recognize incompetence masked by pretty words.
Sam: If you believe this moron’s predictions will come truth in 15 years, take a look at 15 years ago. Of course DEC and IBM will be the ONLY vendors in the market (and maybe Unisys and NCR).
If you stick with 5 years, I doubt than, in 5 years, windows or Linux are able to solve some of the customers problems that Unix and zOS solves today. Since Sun and IBM keep on betting millions on those OSes, My take is that Linux and Windows may or may not grow Market share but, there will be times you’ll need more.
Either way, his comments are plain and simple stupid. Not because someone believes them (so many people believe in so many strange things) but because he’s a VP of a big consulting firm saying that stuff in public
Mary Ellen Lewandowski says:
October 5, 2005 at 5:11 pm
HP’s leadership virtualization based on HP-UX 11i is far ahead of the competition. HP-UX 11i provides customers the ability to sleep at night and provides the best business value for the money.
Here are the announcement details –
http://www.hp.com/hpinfo/newsroom/press/2005/050912xa.html
Here is the real story with even more proof of the longevity of HP-UX 11i – http://h71028.www7.hp.com/ERC/cache/107844-0-0-0-121.html?ERL=true
HP-UX 11i page – http://www.hp.com/products1/unix/operating/index.html
Stiennon says:
October 7, 2005 at 8:20 pm
Right on Sam. While Gartner bashing is fun there are some merits to analysts with a message getting the word out and fostering discussion. Don’t forget that most of Gartner’s client base (80%) are late adaptors. Some of them are still making huge investments in Solaris,and even iSeries mainfram stuff, particularily trading desks and banks. Yet, they may be better off if they looked at Linux for some future applications. At least one already has with over 2,000 Linux servers in their data centers. Late adaptors need assurance from industry analysts that “radical” stuff like Linux is moving to main-stream. This is why “binary” statements from Gartner serve a purpose.
James says:
December 22, 2005 at 10:53 am
The one thing that frustrates me about industry analysts who find faults in the Gartner model is that they never take steps to tell us customers exactly what we need to do to fix it.
Other analysts need to form some notion of a call to action…