Slashdot Mirror
Dell CIO Says "Unix is Dead"
An anonymous reader writes "I thought this might spur some good discussion on this board, including jabs at Dell and MS, which I always enjoy reading. Dell's CIO believes that the end of Unix is here, in fact his opening slide in a recent presentation was "Unix is dead." Specifically, he talked about the savings he claims in moving Dell's Oracle databases from Solaris to Red Hat.
For those of you who came in late, Unix and its workalikes (Linux etc) have grown in use exponentially since 1980.
Professional Wild-Eyed Visionary
Isn't this a little like those trolls that post obituaries on /. for people who aren't dead yet? Anyway, I sort of agree with him, moving to Linux makes the most sense for traditional UNIX vendors that want to keep up with the market.
Anyways, so what?
-Sean
I thought he was including linux in when he said "unix is dead".
I guess not.
Well, I wonder if he's *ever heard* of freebsd or openbsd or netbsd. They are real unix. They won't easily die for a long time.
Liberty.
Then which OSses HAVEN'T reached their end?
BeOS? Dead. (the open source clones/alternatives are far from ready)
Windows? Even 2000 and XP aren't nearly as stable as Unix and are completely non-portable.
Sorry but I don't see any other OS ruling the server market for quite some time.
Large financial organizations are typically *just* moving away from COBOL based apps running on VMS and SCO to Java and C apps running on Solaris on Sun Hardware.
The next Slashdot story will be ready soon, but subscribers can beat the rush and slashdot the links early!
Despite the provocative headline, I don't think Unix can be dead if Linux is alive. Despite the different origins, they are functionally very similar.
Maybe you should have made the headline "Dell CIO Says Closed-Source *n*x is dead". Oh, wait, that might not be quite as good at causing knee-jerk reactions.
Amazing magic tricks
When you spend hundreds of thousands to millions for custom software running on a mainframe, you arent going to be replacing the hardware every year.
Manipulate the moderator system! Mod someone as "overrated" today.
Considering that I've migrated from systems such as NeXT and AIX to Linux-based solutions with very few problems, I'd put forth the assertion that any Linux distribution would qualify as `UNIX' to most lay definitions of the term. I've even taken applications from Oracle/WinNT to Oracle/RedHat with minor issues. Computer operating systems are simply getting better; more commoditizied, which is why Microsoft is afraid of Linux right now. The "UNIX vendors" are still shipping machines, but with Linux installed instead of their "big iron" legacy UNIX systems. I think that he should have said "Operating Systems are Dead" instead -- which is how it should be; the computer should simply get out of our way and let us get jobs done in an efficient manner.
What used to be home-user shops, such as Dell, can now ship high-quality UNIX solutions thanks to Linux and BSD. Quibbling over the proper definition of UNIX seems silly. If it looks like UNIX, acts like UNIX and runs the source found on "legacy" UNIX systems, well, what is it?
The wheel is turning, but the hamster is dead.
Tomorrow's sysadmins and software chiefs are mostly today's CS students. Considering the enormous popularity of Linux with students (for obvious reasons), these new faces will enter the field with much more programming experience and familiarity on Linux than [insert properietary UNIX here]. So, except for very specialized scenarios, I don't think Unix stands a chance.
Just my 2 cents.
-- A humble CS student.
Hmm... Could it be that Dell has an interest in actively killing enterprise-class unix, given that Dell doesn"t manufacture any serious unix hardware. (I know you can installed various flavors of unix on Dell servers and workstations, but Dell has clearly and intentionally linked its own success to Microsoft's.)
This is about as surprising as Microsoft claiming that open source software is crap.
To me, This just smacks of wishful thinking and marketing.
As the boss of Silicon Graphics once famously said: "Linux is the Future of UNIX". UNIX isn't dead - it's just had a major rewrite/cleanup. That's hardly suprising for a 30 year old software package.
The code has changed completely - but the core ideas are exactly as they were back in 1976 when I used UNIX on a PDP-11.
There are more people using UNIX-like OS's now than there have ever been.
www.sjbaker.org
(Yes, Byte lives on in an electronic version - I even subscribe to it - but it's a fading shadow of its former self. It's a lot closer to death than UNIX.)
"Unix is dead, but no one bothered to claim the body" (1986) (from this source.
Of all the pundits out there, Dvorak must have the largest database of being both for and against the same thing; perhaps multiple times. I can even recall him claiming that the Internet was dead. His credibility for me has been zero for several years. I'm amazed anyone reads anything he writes any more.
No one ever had to evacuate a city because the solar panels broke!
I've been in the computer bidness since 1988 and I have heard "UNIX is dead" at least 15 times since then. Every time, it refuses to die.
Here's why Randy Mott, Dell CIO, is wrong:
1: DELL only deals with Intel-based hardware. Intel is cheap-assed commodity based bargain basement garage sale type of junk. Yeah, it works and the speed is increasing more quickly than other architectures, but it's cheap and reliability among different Intel-based systems is inconsistent. Read as: Not big-money mission critical trustworthy.
2: Extremely large database installations running Oracle still choose HP 9000 RISC based machines running HP-UX, Sun machines running Slowlaris, SGI machines running Irix, or IBM machines running AIX. BTW, it's not Linux that isn't trustworthy, it's the chintzy hardware that it runs on.
3: Corporations still want highly reliable iron to run their mission critical processes on. Intel based junk can do it in some cases, but the bigger iron has had better regression testing done on it, and has a better redundancy infrastructure to it, which these companies are willing to pay for. This big iron still runs UNIX, and UNIX still rules the big iron, and rightly so. UNIX -is- however, losing out in the "little iron" and is losing market share from mid-size down, but it's not "dead."
4: Corporations are still willing to pay for all this testing and corporate support for the big iron, if that'll mean big uptime.
5: The only UNIX that is REALLY threatened is the actual AT&T System V that is now owned by SCO-Caldera-SCO again. I used to work for a SCO dealer, and was told by SCO at the time that Unixware 7 was going to revolutionize UNIX on Intel. I told my salesman and managers not to hold their breath waiting for people to line up at the doors to get their copy of SCO Unixware 7. I was right. We sold about three copies of it in two years. We sold ten or twenty times as many of the old Open(Archaic)Server 5.0.x licenses in the same amount of time. Eventually, the new installs became mostly Linux or Winblows, but we only dealt with Intel based junk.
Had Mott qualifed his opinion to mean Intel only, he might be getting close. UNIX isn't dead. I still have clients who would rather run a Sun or HP 9000 any day of the week over an Intel-based machine.
-- I am. Therefore, I think!
Said Schwartz, I don't think businesses are really prepared to trust their mission critical systems to technologies where, if something goes wrong with the open source, nobody is responsible for fixing it and doing all the testing on a timely basis. With Sun, you've got a single throat to choke and we can respond instantly.
This is exactly right. And this is why Dell is very wrong. Them saying Unix is dead is like me saying "Ford is dead" because I personally don't own a Ford. What one company uses is irrelevant. Unix is going to be around for a very long time. Companies don't change platforms willy-nilly, and those that do usually aren't around for very long.
Now, I still greatly prefer UNIX or workalike to NT for any enterprise application. But the extremely expensive, huge geophysical mapping application that I once was the build manager for--which, at the time, was supported on AIX, Solaris, IRIX, and HP/UX--eventually was ported to NT and probably Linux. Also, for example, tons of enterprise-class companies--unwisely, in my opinion--use 2K and IIS and SQLserver.
If you look at what happened in the workstation/server market that UNIX lived within, you'll see that on a market-share basis, UNIX lost an enormous amount of ground to NT/2K. So, the prediction was in a sense accurate but not precise. NT "replaced" what would have otherwise been UNIX installations. However, the overall market increased significantly such that UNIX has managed to remain significant and viable where it still is clearly (and very noticably) superior to NT/2K.
What this reveals is that predictions of these sorts usually have built-in assumptions that are proven false over time. Often, the assumption is of a static environment. This prediction assumed a static market for UNIX and NT where, naturally, the cheaper and sufficiently powerful NT would marginalize UNIX and eventually kill it. If the entire market hadn't dramatically grown and changed in some interesting ways, this would have been true. But why assume that? A more responsible prediction would have been, "NT will replace UNIX in many applications". Which it has.
Yes, but how do you do "less than a fraction"? There is no such thing.
Let's look at the server market:
-IBM has chosen to eventually replace AIX with Linux. This will take for a while, but in a few years >90% of IBMs server will run Linux instead of AIX.
-HP is ditching it's PA-RISC CPU. HP-UX doesn't have any advantages compared to Linux on Intel platform.
-SUN is the only one who invests in own UNIX (Solaris), but it still plans to use Linux in blade servers.
Sure they will be people who will use various BSDs in the future, but from the commercial perscpective UNIX will be dead quite soon (=no new development).
I think most people here missed the more important part of the speech/article, because of the unfortunate headline. The interesting thing about the article was the vision Mott has for IT in general:
a key point in Mott's presentation: CIOs and IT managers need to focus the lion's share of their IT resources on innovation rather than maintenance of the status quo. Otherwise, said Mott, companies and even entire industries will never realize their full potential.
Industries that don't plan for obsolescence will get out of date and they will turn out to be different industries than what they could have been. said Mott.
...
To get out of the rut of obsolescence, Mott recommended a cultural shift. Rather than spending 85 percent of a company's resources on the status quo or keeping the lights on, and 15 percent on development and innovation, the ratio should be turned around.
Now, how doe this square with the responses to the earlier Ask Slashdot about Pointless IT Innovations Considered Harmful? Most people there seemed to agree that a lot of the "upgrade cycle" was pointless, but here we have someone from Dell claiming that we should spend even more on useless upgrades becuase the industry depends on it. Hmm.
Linux IS Unix. Yes, I know it has no "standard" Unix code. Yes, I know Linus Torvalds doesn't have a license to call Linux a Unix. So what?
So what makes something Unix? All of them have some differences, but there are a number of commonalities. You'd never mistake an MS operating system for a Linux system, for example. Though it's not correct to say so in some circles, I say that Unix is as Unix does. If it looks like Unix, and more importantly, ACTS like Unix, it's a Unix.
Basically, if it uses most of the standard Unix commands, and it uses one of the Unix shells (Bash, Korn, etc), and the OS code looks like a Unix (Kernel, Shell, Window system, etc), its a Unix. Even the Kernel isn't as thorough a guide now, as there are enough signifigant differences in "real" Unix systems to make this factor somewhat iffy (monolithic kernels vs. microkernels, for instance).
So to say that Unix is dead because Linux is replacing many traditional Unix systems seems a little disingenous. Just my 2 cents on the issue...
Life is hard, and the world is cruel
Why would he say anything else? Considering that he's referring to commercial/proprietary UNIX systems, it obviously is in his best interest to make such a statement as the commercial/proprietary UNIX systems out there are usually tied to a specific type of hardware. Plus, Dell generally seems to target consumers and end users in corporations, not server rooms.
Although Pedants will remind us that "Unix" is copyrighted, in practical terms, Linux is as much Unix as Solaris.
The claim that the fellow is moving "away from unix" is ridiculous and misleading.
Hmm... Lets see... a huge distributor of PC hardware makes a decree that UNIX is dead.
Many of us know that the PC platform is unpopular as a hardware environment for commercial unices.
Could it be that Dell's CIO is really saying "The hardware that runs UNIX, such as SPARC, for example, is dead. Buy a PC with Windows or Linux and never have to worry about having obsolete hardware."
That's what I read between the lines anyway.
Saying "Linux spends almost no money" in R&D is incorrect. Or rather, vacuously correct. There is no "Linux" entity, therefore it cannot itself spend money.
The correct statement would be "The Linux community's investment in R&D is impossible to estimate in monetary terms, but is likely to be less than Sun's".
Just because the research effort is not centralized, paid for from a single source, and is difficult to account for, doesn't mean that "the community" doesn't still spend a lot of man-hours, hardware and software in R&D.
That includes individuals contributing their time and resources for free. But it also includes companies and corporations invest money, to earn more money: from Linux distributors to corporations like IBM (whose investment in Linux has a lot to do with R&D).
I'm not saying that the Linux approach is better, but to pretend that just because no single company is paying the bills there was no investment is incorrect.
Freedom is the freedom to say 2+2=4, everything else follows...
I find your attitude curious. Dell makes decent machines, and they're essentially the first choice most people look at when buying a system, inlcuding people who do have the capacity to build their own systems from PriceWatch parts.
Ecce Europa - Web Design for Business
Said Schwartz, "I don't think businesses are really prepared to trust their mission critical systems to technologies where, if something goes wrong with the open source, nobody is responsible for fixing it and doing all the testing on a timely basis. With Sun, you've got a single throat to choke and we can respond instantly."
The thing is, that level of support comes with support contracts, not with simple purchases. Once you start making the case that the superiority of your OS is based on how you will respond to support contracts, however, you've gone pretty far down the slippery slope, IMHO. Perhaps it is impossible for a Linux distro (or some third party) to ever offer that same level of support, but I wouldn't bet money on that. What will Schwartz say in Sun's defense then? Of course, he may be working for a commercial Linux distro by that time, and will have no interest in trying to come up with a defense for Sun, anymore. Who knows?
The thing is, I don't find that "something goes wrong" with the kind of regularity that Schwartz seems to fear. Most of the time when we have Sun or Dell out here to service a server, it is to replace a hard drive in an array. The service contract is basically a way to avoid having replacement parts around for mission-critical systems. Is that enough reason to go ahead and buy the extended warranty on your OS when you make your purchase? I guess businesses will continue to decide that over time, as Dell has.
The same is going to happen with databases. While there doesn't seem to be a good open source, distributed, redundant database for Linux yet, many people are already effectively building such databases out of MySQL. Yes, MySQL. You see, not only can the hardware be less than stellar in redundant, distributed systems, the software can be as well. And if you like a COTS solution for Linux, IBM already offers it.
Scalable databases will become as simple as buying a bunch of PCs with large disks, plugging them into a high-speed switch, and network-booting them. If you need more power or one breaks or goes down, you just plug in another one.
In the end, combining lots of redundant, cheap units gives you much better reliability for less money than the overly expensive and overly engineered "reliable servers". Because, no matter how reliable a single server may be, sooner or later it is going to break, even if it just because someone spills a comp of coffee into it. And the solution to that people are using right now? They are buying two very expensive high-end servers and use one as a hot standby.
This is no more than a cheap sensationalism to sell more Dell servers by dividing and conquering the Unix and Linux community, and would also strengthen MS in the high end server market which is still dominated by Unix.
By it's own admission, Dell profits from other people's R&D budget. This is one of the richest company in the computer industry with no technology other than cheap box making skills and makes zero contribution to the world. It's well on the way to become an MS-like monster playing every trick in the book to kill its rivals.
I for one can't bear the thought of a world full of ugly Dell boxes with dirty Windows. For the sake of our industry, we need the innovations of Apple, Sun, IBM and many others, so let's boycott Dell boxes - they are not even cheaper anymore.
Except where functionality is actually removed. E.g. Stallman insists that man pages are obsolete and refuses to support them, which is incredibly wrongheaded.
I'll second that.
Info always was a pain in the butt - especially for those who don't use emacs as their preferred editor. And now that HTML (and its augmentations) is here and browsers are essentially universally available, info (which never achieved the user penetration of man) is itself at least as "obsolete". Considerably more so, in fact, precicely due to that lack of penetration.
Sometimes an adequate standard is better than a "better" multi-standard solution. Books, for instance, are not obsolete (even if clay tablets have been depreciated for a while.) Man is just a convenient online set of loose-leaf notebooks (suitable for hardcopying for those times a spare square-yard of screen isn't handy).
Needing a mix of tools to read the minimal subset of manuals is so broken it hurts my head just to think about it. The "man is obsolete and gnu won't support it" case of Not Invented Here is one of the biggest impediments to general conversion from proprietary software to Gnu offerings.
Fortunately there are info-to-HTML translators. Unfortunately, I have yet to meet one that conveniently ports all the info functionality into the browser environment - which is a problem, since the "info" manuals were written assuming it and pretty much require it for effective use.
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
While I run linux on my desktop at home and on the desktops at work, I have been a Solaris admin for a decade now. I watch the innovations fairly closely, but Linux just seems to keep falling short on serveral higher-end issues. Linux just feels too much like something made from a "hacker in a basement" with no thought to enterprise issues.
/etc and other file systems, along with whole subsystems moving/disappearing/being replaced. If we ran enterprise legacy apps on a Linux box, we would not dare upgrade it, ever.
Here's my (admittedly anecdotal) experience:
* Intel hardware just does not have the bandwidth to keep up on general business apps. We do quite a bit of batch processing under Oracle, churning through tens of gigabytes of data per day, and every time I test the same apps under Oracle or MySQL or Postgres on a Linux box, the machine just endlessly grinds away, even when the Intel gear is running radically higher clock rates.
* Linux seems to collapse under higher processing loads. We routinely get our V880 to load averages of 12 or more, with 1000+ processes, and all it does is get progressively more sluggish. I have yet to see a Lintel machine stay above a load average of 4 without falling apart.
* Linux apps/developers seem to be enterprise-hostile (or at least unaware). When we switched away from Openwindows, we tried KDE, but had trouble setting a common launch menu for all of our users (attached to the Sun via X terminals). Worse, some logins had to stay Openwindows (for a variety of reasons) and there seemed to be absolutely no way to keep openwin-menu and the binary KDE menus in sync. A message to the KDE developers asking for advice resulted in childish responses about our organization living in the past. We went with Ximian Gnome for the menuing support. The sick thing is that the most recent Ximian Gnome (1.4?) gets confused when the same person logs in from multiple terminals. A message to the Gnome developers asking for advice yielded no response. There is still no graceful way to systemwide disable xlock. 50 users running xlock on a central machine eats up the CPU and floods the net. We have a hack that we do to keep xlock off of the machines, but it gets overwritten by each Gnome upgrade.
* Solaris, and I presume AIX and HP-UX, are very upgrade friendly. We have come from Solaris 2.2 to Solaris 8 and swapped the server hardware twice over the past 10 years with no disruption to our 1000+ custom legacy applications. The worst problem we ever faced in an upgrade is that we run a non-standard sendmail and Solaris will occasionally stomp on it in an upgrade. Each new Linux upgrade seems to bring new layouts to
Hopefully Linux will grow up and be ready for larger installations. I am still waiting.
It seems most appropriate here...
"Don't advertise your own stupidity.'
Unix is dead, 640KB is enough for everyone.. It's a new economy... etc
Just don't.
Clear then this article was little more than an argument between Dell and Sun over Dells switch from Solaris to Linux. How this spells the end for the UNIX model is quite beyond me.
-- "Can't sleep, clowns will eat me!"