Linux only has a small percentage of the computing market, however Microsoft already considers it a major competition as the open source OS steals the hearts of many users. Following the hard numbers though, Microsoft also increases its market share on both server and desktop space with time. The only logical explanation is that Linux steals quite a market share from the traditional UNIX providers (SCO, Sun, SGI, HP, IBM). But only Sun seems to truly be in a real Linux trouble, as it is the one with a resistance to Linux integration to its full product range.Many consider Linux as the natural evolution over Unix. It is a re-implementation, largely compatible and while it doesn’t have all the features found on high-end UNIX OSes, it contains others that can’t be found in these propriety, commercial Unices. At the peak of Linux’s hype in 1999 and 2000, the main Unix providers re-arranged their strategy to include Linux, as they found that it just… sells. Except the lower price and the OSS mind share involved, there is nothing that Solaris, IRIX or AIX can’t do that Linux can. But the OSS-hype drive is strong enough to re-define the high-end server market.
IBM is selling Linux most of the time running under runtime engines rather than running it as the baseline OS. Linux beefed up the sales for IBM and in fact the company is very pleased of the market performance it gets by the the momentum it generates. AIX is still there, running Linux under virtualization, but where that leaves AIX as an overall useful OS running without the need of Linux?
SGI is selling either Linux Itanium servers or IRIX on MIPS, but their focus lately is totally on Linux. So where does this leave IRIX? We haven’t seen a major new version of IRIX for years now.
Sun is somewhere in the middle of the whole situation. Publicly they try to not stir the OSS community and place themselves as a “competition”, but in reality they still embrace Solaris more than Linux as their main product. Sun is to offer Red Hat Linux in the “Project Mad Hatter”, which aims to offer cheap LX/60 machines to their existing customers in the case of these customers say something like: “hey, these SPARC machines of yours are expensive for what we need, we will take a look to cheap x86 machines or even to Microsoft’s solutions”. In other words, Sun’s strategy on Linux is a backup one. But how much life Solaris still has under the competition of cheap Linux servers?
HP is only using HP-UX as a very specialized solution, mostly in the medical industry. HP-UX’s future (together with Solaris’) will probably be the one with more prospect from what we can see in the market today. But possibly not for long. HP already has made its Linux move, they already sell Linux machines and servers, while their other Unix, Compaq’s Tru64, is now getting integrated into HP-UX itself, so we won’t be seeing more of it in the future as an individual solution.
SCO’s strategy is well known to everyone. After Caldera bought SCO a few years ago, they grew weary of Linux and it is now the only big Unix commercial company that holds on to its original product and in fact, fight Linux via various ways. These “various” ways though they don’t help SCO to market their product in a favorable way. Customers are already turned away from SCO because of their recent behavior. If Linux truly replaces Unix with time, SCO Unix might be the first of the Unices to go down.
So what’s Linux’s role to the Unix timeline? Is it the natural evolution of the UNIX architecture that comes as a storm and replaces the old propriety products, or is it a temporary blip to the Unix history? Or is it a parallel technology? Is Linux a compliment solution or is it a replacement solution? From what we see so far, Linux moves towards replacing all Unix as we know it, and thankfully, these Unix companies realize it and embrace the change as fast as they can (except SCO of course, and Sun). Microsoft realizes these changes as well and without doubt doesn’t feel comfortable for them, as so far, Unix was a pretty dichotomized market. With an un-forked Linux kernel advancing in a very fast pace and getting developed by all these companies at once, is surely a huge competition for Microsoft.
The reason I wrote this editorial was because of my own “romantism” towards operating systems. One part of me believes that having a single OS running on all devices conceivable is great means for interoperability (think “Star Trek” and how they connect alien devices to their own with ease), but the other, “osnews” part of me, loves to see more and more operating systems and architectures on the plate. Seeing old traditional Unices fading away with time, or at least losing their glorious role every day to Linux or Windows, truly saddens me.
Why must everyone switch to Linux? What features does Linux have that real UNIX OS’s don’t?
As this article correctly points out, companies may be adopting Linux, but only by running on top of an existing system. Why is this? Linux is missing critical features needed to run large computer systems. Some examples are checkpointing, hotswapable CPUs, various specialty hardware, stability testing of the large UNIX OSs.
Interestingly, Linux users seesm to love to badmouth Sun. This is somewhat ironic because Sun was behind some of important aspects of the GNOME desktop. For instance, they really started the push for fixing up the GNOME UI, and recently added voice synthesis to GNOME for blind users. They have not adopted Linux as an OS because there is no need. Linux cannot do what Solaris can do (hot swap almost every component of a server for instance), and Solaris can do pretty much everything Linux can do.
One of the strengths of UNIX has always been well published standards. This allows different implementations to be used. There is no reason at all that everyone should use only Linux as the kernel. Nearly all the software for Linux can be compiled for other systems as well.
Because it is so similiar. It is very easy to port your apps from one version of Unix to another. Much easier than say migrating to NT. The traditional Unix vendors will have to value add to their products for them to stay afloat. But those old Unixes will die against Linux.
>, but only by running on top of an existing system.
Actually, only IBM pretty much does that from the big 5 of the Unix market, AFAIK. The rest use Linux as a stand alone OS, replacement or complimentary to their Unix solutions.
First of all good article.
This change that is going on is just natural evolution taking hold. This has been the standard for many centuries. Take for instance through time there has been many superpowers that have come and gone. Ancient Greece, the Roman Empire, the Spanish Empire, the British Commonwealth, and today the United States of America. All these had their glory days, and then have withered away because people change.
This approach is the same in the IT industry. UNIX is like Ancient Greece. It is now dying away to be replaced by the Roman Empire (eg Microsoft), eventually that will fall away to be replaced by another superpower. Like I said before, it is just natural evolution taking place. We don’t know when the next big change will happen, but when it does we will be ready for it.
I’d say SCO is in deep, deep shite.
They won’t be missed.
Sun will be all right, they have some quality about their organisation. Sun hackers do Gnome contributions for example. Selling very popular software is not a viable business model (OS’s, Office suites, servers, browswers whatever) Sun have nice hardware and expertise to sell. I don’t know what Microsoft are going to sell yet.
I love Sun, mostly because of the things they’ve given the computing world (Java, NFS, etc, ad nauseam). However, Linux can do one thing Solaris can’t: run at its full potential on x86 hardware. This is why Sun is losing market share; *SPARC machines are slowly being irrelevanted.
Originally, the strength of UNIX was its relatively open development model. When that started going away, UNIX first went a little up, then back down because of corporate fragmentation–where MS gained its beachhead. I think Linux’s gradual ascendance is because it returns to the old, open model.
>I’d say SCO is in deep, deep shite.
That’s true, but what they did to get in these deep shite, happened with their own will. SCO left the Linux market under their own decision, Sun is trying to play it “double”.
It’s not only sad, but it means there will probably be less innovation. While both Microsoft and the open source community have produced real innovations, they are perhaps better known for being fast followers (for example in GUI, browsers, bytecode computing). Companies like Apple and Sun will have to kept around just so the eventual big two will have something to co-opt, heh heh.
But another platform seems to be opening up around handheld wireless computing, so the cycle begins again.
—ANNNOUNCEMENT—
The Linux kernel developers have decided that Linux should take a different route. As per that goal, the new focus of Linux development will be LxHWA (Linux for Handheld/Wireless Appliances). Further development of the Linux kernel will be slowed down.
Thank you for your support of Linux.
That was fast!
Yeah, that’s me, Memorial Day and nothing better to do than hang around here.
Well, I would like to see Apple and SUN merge into one company. No laughing 😉 Just consider this, an Ultra Sparc III processor in an eMac running an OS with a Solaris core and the MacOS X GUI running on top of it. It would be the ultimate nightmare for Microsoft, the combining of a rocksolid UNIX server company with the market savvy cusumer focused culture of Apple.
I dunno. I think the Darwin kernel is perfectly satisfactory–how many Mac users are gonna use ANY of the capabilities of the Solaris kernel? It would also mean significant and useless overhead.
Also: UltraSparc III? On a desktop?
New! EMac SPARCqua edition–only $15000!
>> but the other, “osnews”, part of me loves to see more and more operating systems and architectures on the plate.
>>
I agree completely, but the real problem in this respect is windows, not linux replacing unix. Windows already runs ninety-something % of OS-powered devices!!!! It thoroughly dominates the desktop, has a huge and growing market on the server, and is creeping into everything from Palmspilots to cell phones to BMW dash-boards! It even wants to be in your toilet! Compared to windows, Linux is just a drop in the ocean.
If, like me, you like to see more operating systems and architectures, support linux. Why? Because Linux is open, is available to IBM, Microsoft, Sun, Hp, or whoever cares, and is not tied to any particular monopoly or corporate agenda.
Should linux replace Unixes? I would hope so! It would be a great way for Unix outfits to pull their resources together, instead of fighting each other in ways that only benefit abusive monopolies like MS.
Also, linux has a pretty good chance on the desktop, so every company supporting linux might eventually help create a more open desktop environment.
Interestingly, Linux users seesm to love to badmouth Sun. This is somewhat ironic because Sun was behind some of important aspects of the GNOME desktop.
… and look what’s happened to gn0me. It’s the biggest joke ever. Especially Gnome 2.0. Is gnome a CDE replacement? Nope.
While solaris does have its technical merits, and Sun have contributed quite a bit to the OS community. If they keep on doing what the do, that is, bash M$ and little else – they’re goin’ down, while Linux is catching up.
>> Interestingly, Linux users seesm
>> to love to badmouth Sun.
I love SUN, or used to love them. They used to be one of innovative companies around. Thesedays though, about the only thing they do is REACT to microsoft, and maybe IBM. That’s not the way to run a company, and that’s a lot of people bad-mouth Sun.
Solaris? Great OS, but hey, I walked into a huge ISP this last weekend, and guess what? They are replacing all the Sparc machines with Linux on Intel. Price-wise, Sun hardware has become a looser to Intel. They need to adjust to that reality, and a Linux strategy is the quickest way to do it.
I’ve read Eugenia’s article and think it deals with a lot of different aspects.
1) Linux is portrayed not as the only alternative to proprietary Unices but as the standard bearer of Open Source. That “Linux hype” is a media creation. I guess most IT reporters know about other free *nix.
2) Open Source software stresses out the intrinsic weakness of IBM, Sun,… : these corporations have developed quite an expertise throughout the years. But, when it comes to their workforce, they are no match to the number of volunteers available to Linux or *BSD : these coders strive to freely improve every single aspect of their preferred OS.
3) As a consequence of the second aspect, businesses take a look at free software because it is cheap and is maturing fast.
4) Why should businesses still believe that only proprietary vendors can meet their expectations when there is ample proof of the contrary (Yahoo running on FreeBSD, NASA using NetBSD, financial corporations using Linux, etc.) ?
5) Free software make good use of old hardware, particularly the x86 platform. This is another way Unix vendors have goofed : they put the emphasis on corporations while Intel and AMD were busy catering to countless home users. HP and SGI are so tired of their own CPUs that they have become Itanium resellers. Sun produces computers which run without interruption for days but are relegated to the workstation and server market.
6) Interoperability is not related to the use of a single OS like Linux. It is guaranteed by the ability for an OS to follow specifications and standards which are available to everyone. After all, I can use a *BSD as an AppleTalk or Samba server as well as I would with Linux.
7) Apple is also a Unix vendor, yet its existence isn’t threatened by Open Source software. Too bad it has a market share of less than 10 %.
Of course, I may be completely wrong.
>Apple is also a Unix vendor, yet its existence isn’t threatened by Open Source software.
Apple is NOT a traditional Unix vendor. Apple just now emerges in the server market, their main strategy is completely workstation-oriented, therefore it has nothing to do with the market of the big 5. Linux can’t compete in the workstation/destkop arena yet, therefore, for now, Apple is not part of the “endagered species” as the other Unix vendors are.
>That “Linux hype” is a media creation. I guess most IT reporters know about other free *nix.
Yes, they know about BSD. But most of them STILL prefer Linux. Why? Because it SELLS. This is what the hype is all about,.
…
me included.
Having present that Windows is a workstaion/desktop which
is starting to get server computers market, Linux would
only increase its (market?) share by turning itself into
a decent desktop system (Monsieur de La Palisse dixit
.
However, that isn’t hapening since kernel 2.2.10 and
the comunity of developers of Linux can’t think has a
bussiness strategy board of directors >> Implement and
finish usability and maintenance aspects first !
like software installation, setup (Control Panel) that
are standard and easy for any user, directory layout, and
sooo many important issues. Unless the owner (Linus
Torvalds) claims something about this (say, binary can only
have this and this to interact with kernel, directory layout
can only be this and this, etc). In pratctice, this might
impossible, but … I would love to see Linux on the same
level of polishment of Windowsor Apple OS.
Maybe Windows will turn out as a safe and decent modern
server platform after all, maybe even to the mainframes
or clusters.
http://www.microsoft.com/windows2000/technologies/clustering/defaul…
I mention this since I found many astronomy and medical
applications in Europe to be written for the windows
platform beneffiting which will use this cluster power.
There is a place for every flavor of Unix and Unix-like operating systems. Looking at it from a Corporate perspective (which, thankfully I no longer belong to), and a grateful last 10 years as an SysAdmin, I’d say “If it ain’t broke, don’t fix it”.
It’s like this constant migrate/upgrade cycle, it doesn’t really matter which OS you use/admin, It’d be nice to see each OS stand on it’s own two feet. Each trying to copy the other, except where required by marketplace… they’re innovation is lacking…across the board. Whew!
Like a constant game of catch-up, no side is really a winner. The reason I say these things, is that …when it comes to computers, diversity is an ally. Single-point of failure is a bad thing and the “I am glad we don’t have a XYZ-only shop” is a good thing.
Just venting.
=)
.
> Why must everyone switch to Linux? What features does Linux have that real UNIX OS’s don’t?
Linux is probably the best server OS on x86 (besides the *BSD’s…)
x86 hardware itself has grown in mindshare with a lot of companies and people.. this growth can not only be attributed to Linux/BSD’s, but also to the growth of Windows as an server OS.
> As this article correctly points out, companies may be adopting Linux, but only by running on top of an existing system. Why is this? Linux is missing critical features needed to run large computer systems. Some examples are checkpointing, hotswapable CPUs, various specialty hardware, stability testing of the large UNIX OSs.
Yes, adoption of Linux goes on in more of the workstation and new server level.
>Interestingly, Linux users seesm to love to badmouth Sun. This is somewhat ironic because Sun was behind some of important aspects of the GNOME desktop. For instance, they really started the push for fixing up the GNOME UI, and recently added voice synthesis to GNOME for blind users. They have not adopted Linux as an OS because there is no need. Linux cannot do what Solaris can do (hot swap almost every component of a server for instance), and Solaris can do pretty much everything Linux can do.
Uhm, how is GNOME tied to Linux? I’ve used Linux for nearly five years now, and I’ve barely ever used GNOME (I prefer fluxbox and KDE, neither of which are tied to Linux or Solaris, or Windows even.)
> One of the strengths of UNIX has always been well published standards. This allows different implementations to be used. There is no reason at all that everyone should use only Linux as the kernel. Nearly all the software for Linux can be compiled for other systems as well.
Yes, but often unfortunatly, there is a lag period for non-Linux systems… a recent case in my experience was trying to get gcc 3.3 installed on a HP/UX 9 server… I only did with much hassle and installing several sets of patches.. it worked pretty much out of the box on my gentoo/linux box a week earlier, however. That server, ironically, is primarily for developers. If we didn’t have a lot of legacy custom made close source apps from defunct companies, we might have even gave Linux a shot.
I pretty much lost all respect for Sun. I do not know if any of you watched the Sun Webcast with Scott McNealy and Larry Ellison, but in it Scott takes several cheap shots at Linux and even feeds some of SCO’s press. I like Sun products but Scott is becoming another antagonist. In my opinion people need to STFU until SCO decides to make available any evidence to substantiate their claims. Because IF SCO is lying and just doing this to cause FUD then several people will be eating their words.
{ >Apple is also a Unix vendor, yet its existence isn’t threatened by Open Source software.
Apple is NOT a traditional Unix vendor. Apple just now emerges in the server market, their main strategy is completely workstation-oriented, therefore it has nothing to do with the market of the big 5. Linux can’t compete in the workstation/destkop arena yet, therefore, for now, Apple is not part of the “endagered species” as the other Unix vendors are. }
Apple also sells to a very limited audience, it is a niche market as I have stated before Apple has its customer base, Linux has its customer base. Even after their slight emergence in the Server market I have yet to see any substantial flocking to a Mac server such as the Xserve or any of their other products. The server market is littered with Linux, Windows and some UNIX variants, it is hard to win any customers in that Area. Mac will continue to be popular among Mac fans and you may see some people buy into the hype and switch but you wont see them go in droves and you wont see Apple gain really significant marketshare in any arena.
{ >That “Linux hype” is a media creation. I guess most IT reporters know about other free *nix.
Yes, they know about BSD. But most of them STILL prefer Linux. Why? Because it SELLS. This is what the hype is all about,. }
Linux is more popular because of Ease of Use and system design. People like automated installation and do not wish to sit around and have to figure out a text installer, they want something idiot proof and some Linux distributions offer it. BSD has its strong points and has its advantages but the disadvantages outweigh its advantages.
I myself am quite fascinated by the other architectures asides from x86. I guess I like Linux well enough, but I’m more of a Unix enthusiast than a Linux one. Lately I’d come to the same sort of conclusion about Linux, with respect to the longevity of the big iron Unix lineup.
MS may have 95%+ of the market but they aren’t the threat to Unix diversity or Unix itself. Windows is windows, I couldn’t care less, but Linux is, to take the usual star trek theme, like the borg. Slowly but surely it’s absorbing the rest of the market and taking over whereever the other Unices are moved out, be that SGI shops moving to x86 render farms or Alpha clusters for (more) x86 ones. The Unix vendors have no weapon against it that differentiates themselves for long. I don’t really think that by and large companies suddenly say ‘let’s move to Linux’ from another Unix line, but when their flagbearer trips up (SGI, maybe Sun) or buries them (good old HP) it’s a good enough alternative, and not a business decision I can blame them for, considering. But in so far as I can see it’s a trend from dedicated systems that excel (for their time) to generalised systems that will just do, the watermark of x86 I’d add.
To place this admitted ramble in context, I personally have x86, SPARC, Alpha, PA-RISC, and SGI/MIPS computers (and if you count PDA’s, ARM too). Running Linux on each of these widely different architechtures is near exactly the same. For me personally that is entirely besides the point – you might as well just get an old pentium for all their actual performance but that’s not of course why I bought them. The fun part is in exploring what these systems individually do well, and they can only do that with their “own” Unix running on them. The death of these systems heralds the death of elegant solutions to computing problems, but that’s not all Linux’s fault (and at that, for being too good), but it’s part of it.
I do not dislike Linux, just what it’s going to take part in doing to make the not-windows world as homogeneous as windows itself is.
When thinking about it very broadly, I believe market share will change but I think only UnixWare and OpenServer are poised to go down the same path as CP/M, which is a shame because there were a few things I liked about UnixWare. My first Linux experience was with the fledgling Caldera OpenLinux, DR-DOS and WebSpyder were rather cool too and so was the company in those days. It’s amazing that they have chosen a ten guage shotgun and 0.0 buckshot to blast away at their own feet. The “hoist by owm petard” analogy at The Register a few days ago was sadly perfect.
The article raises the question of whether Linux will be just a blip. I think not.
The Unix companies out there offer their own flavor on own hardware trying to dominate but going nowhere. No customer wants to get locked – especially in these times.
Linux is the standard that will rise from the ashes because it runs on a standard – i386. Also Windows runs on i386. And there are 2 suppliers of i386 – Intel and AMD – to keep prices down. The latter heavily backed by IBM. Strange that none of the big asian companies have started a 3rd incarnation yet.
The key in the movement is not Linus. If not him then someone else would have started Unix on i386. The key is the GPL by Stallman. That’s what’s keeping it together.
What happens when Linus does not want to manage it anymore (allthough a successor already has been appointed)? When it grows so big that it is serious business? Will open source developers around the globe then be commiting changes to IBM as maintainer? Who’s version will the customer pick – the one maintained by IBM or kernel.org?
Are there any UNIX vendors that sell UNIX that don’t also sell hardware on which to run it? If not, how much does it matter whether UNIX market share is moving over to Linux? After all, UNIX would just be extra value added to the purchase.
To answer my own first question, I just thought of Sun, who sells an x86 version of Solaris without also selling x86 hardware. And that’s probably why they don’t like Linux.
What features does Linux have that real UNIX OS’s don’t?
It runs on dirt cheap intel boxes and has boatloads of commercial hardware and software support.
In short, if business have applications that can be horizontally scaled onto lots of small, cheap intel boxes running Linux, then they are migrating in droves.
I’m not quite sure why this is news, though. Everyone has known Linux is taking all its marketshare from the commercial Unixes for years. It’s made barely a scratch in Windows marketshare for this very reason.
Basically, the only reasons you’d be running a proprietry Unix system today are:
a) Software and/or hardware lock-in (eg: you need Solaris for some software your business depends on, or you want to use hardware that is only certified with, say, big Sun Enterprise machines).
b) You have a need that is better served by (say) a single SunFire 15000 than by a few hundred intel machines running Linux.
The most common reason is (a). Category (b) is small and getting smaller every day.
I have never understood why Sun and Solaris is so popular. I have nothing against Solaris, but I really like HP-UX and AIX much better. It looks like HP and IBM are in a good position with Linux while Sun talk much about Linux but does not do much.
SunOS was a superbly good implementation of UNIX. Solaris took a long time to match the reliability of SunOS. And it has never matched the hands-on tactile responsiveness of SunOS.
Sun’s mission with Solaris was to create an OS for their mainframe-killer computers. And for the most part, Sun has succeeded.
Thus today’s Solaris a giant OS most suited for giant systems. It is no longer the lean, mean and reliable workstation / small server OS that SunOS used to be. It’s no longer a fun OS, but a heavy duty consulting-enabled corporate OS.
As most of the innovation is on the small systems, Solaris is left with no evolutionary path forward except to become more and more of a mainframe OS.
Linux is the OS for the new small scale systems. While Linux has also been adapted to run on large systems, at heart it remains an OS for small servers and workstations. With Linux available to many people on low-end hardware, an incredible amount of development can take place in the Linux community.
It is much like the tale of mammals vs. dinosaurs. Sun and Solaris are dinosaurs. The company and the products are mostly large and heavy, slow and ponderous. It’s a headache to even deal with them. Their salespeople are brutally pushy and disgustingly clubby. Yuck.
Tell me ONE really cool thing that’s happened on Solaris in the past couple years. That’s pretty much the crux of the story.
Maybe Sun can get together a bunch of their big machines in a cluster and call it Jurassic Park.
IMHO, if Sun can hang on long enough to deliver Solaris 10, I think it will be something like this:
All/most user-space programs, libraries, etc — the user “environment” if you will — will be common (and LSB- compliant – this is already a stated goal for Solaris) between their Linux and Solaris products.
The kernel will be the differentiator: you need super-SMP, hot-swap, etc? Plug in the Solaris kernel on SPARC hardware. You want lower cost, 1-4 way (server or desktop)? Plug in the Linux kernel on x86 or SPARC. But most everything else will be the same between the 2 versions.
Then they will have a “complete” offering that becomes mostly hardware independent (will anyone really try to do a 100-way x86 server?) and yields a common “look and feel” to the users and the developers (one common API for 98% of everything; specializations — as they will always exist for any vendor — for the bleeding-edge massive 100s of CPU systems).
And they’ll have the complete range of hardware — from 1-way x86 servers (like the new V60) and desktops (“Mad Hatter”) to SunFire F15K and beyond “big iron” boxes. With N1 (as it develops) to help manage and integrate it.
If they can just hang on…
yes, I could see an Ultra Sparc III based Mac. 90% of the cost associated with UltraSparc’s pricing is due to two things, firstly the lack of volume/economies of scale which can be corrected via the large number of Macs sold each year, and secondly, the pricing of UltraSparc III chips have a truck load of cache on it. Scale the cache down to 512K for desktop use and you won’t have a problem.
SUN can compete with the Lintel (Linux + Intel), however, they need to start agressively cost cutting, that includes outsourcing all their production and spread the production of all their chips to UMC and TSMC rather than simply assigning a token about to UMC which they are doing right now.
very interesting article,
but a couple of things to note for sun.. no one who owns a high end sparc would be installing linux on it.. ultimately linux just couldnt provide what suns solutions do provide.
dont get me wrong i love some of sun’s stuff. and ultimately hope they dont go under. but suns main selling point has been the ultra sparc 64 bit processors with intel and amd both jumping to 64 bit albeit many years later.. this means sun has to be offering better to compete are there actually any 128 bit processors in existence ? is it actually even possible. or they really need to wisen up and start selling their hardware cheaper.. go onto their site and make your ideal desktop i dont call $50,000 affordable..
why would u play $800 for a dvd rom drive when it only costs £40 for intel/amd hardware..
why linux has ben so well selling well put it this way the x86 hardware they have shown ultimately to be stable enough. x86 servers with uptimes over a year again *bsd is capable of this.. but bsd gpl application wise is always lagging..
about 5 years ago.. companies had choice of either using ms windows. which was and still is notoriously unstable.. with having to reboot to do the smallest things. or nix which required deep pockets but the result was damn solid performance.. but now they have the middle ground.. affordable nix capable of running on nearly any hardware and ultra stable.. which would u buy ?
Once I did hold hopes that SysVRx – SCO Unixware – could re-establish itself as a competitor to Linux and the *BSD Unix lineage. I thought if Caldera could offer its Unixware and OpenServer clients the SysVRx source code under a version of the BSD license, with support contracts and reductions in support charges for successfully debugging problems, perhaps it could survive.
You know, you do something positive with the source code to show you’re an active member of the community, and you get rebates on the membership charges, that sort of thing. I think it would’ve worked.
Now of course SCO – formerly Caldrea – have chosen Russian Roulette with a fully loaded, well maintained Uzi
http://yro.slashdot.org/yro/03/05/26/1648232.shtml?tid=106&tid=123&…
… Sun has pulled back from that sort of brink, with its begrudging adoption of Linux, and its continued development of Solaris x86.
None of which affects the gradual growth of completely different architectures, such as Syllable, VSTa, OpenBeOS, etc.
I think the completely new software like Syllable, etc, merits watching. I don’t think the original Unix is worth anything now.
I agree, I personally can’t see anyone installing Linux on a beautiful Starfire. Solaris has the most scalable kernel in the operating system market, the only one that comes even a close second is HP-UX.
Solaris is improving and from what I have heard, Solaris 10 will be a ripper of an operating system. Yes, Java has its issues just like .NET, and sure, I hear Slowaris jokes and heck, I even know a couple of SUN engineers in New Zealand who called it that, however, as they said to me, once up and running it is a little speed demon. Solaris only got that nasty reputation for its slow booting. IMHO, I haven’t come across it any way. Maybe my expectations are lower than some peoples?
As for the over all progress of SUN, I’ve had a look at their balance sheet and if you take off the one offs, they’ve made a decent profit for the 9months ending March 2003 with software and services growing. As for the outlook, anyone who thinks that SUN will reach the dizzy hights of the dot-com boom, sorry to let you down, they will never achieve that sort of growth. I could see however, the UNIX market grow in the next 7 years to clost to $22-$23billion, however, anyone assuming that there will be the same growth rate as before, well, prepared to be disappointed.
I think some considerations in this article are wrong.
First of all Linux does not have now a small percentage (in the server market).
Second it’s first a re-implementation and then an evolution and of all the Unix O.S. it has highest trend of development, the largest and differentiated team of developers (and I am only considering the kernel) and it’s the more customizable OS solution today on the market (from wristwatch to enterprise mainframe). Also in the most of the cases Linux (server) has all that you need for implementing IT services.
Besides other “classical” aspects like scalability, on Intel systems (where Linux will be competitive in less than a year: next 2.6 kernel), there is a fundamental feature that’s very important in the IT today infrastructures: the integrability.
No other OS offer, in the server arena, the same kind of versatility of Linux. All the best open source (and now also commercial) server system applications (web, app.server, mail, etc.) have Linux has the reference platform.
More over Linux will be soon really an alternative also on the desktop.
Consider also that the x86 (and the evolutions) platform now is now at the same performance level of the RISC counterpart at a fraction of the cost.
So why someone should have a long term strategy on other (expensive) solutions (O.S.)?
Work in a company with an MS IT shop (I am a research chemist not an IT professional). A year ago I told our CEO as part of my Linux evangelism that you could run Oracle on Linux – He said “I didn’t know that”. Today our first onsite Oracle deployment is running on RHAS. If our IT people had been allowed to carry on in their MS monoculture way they would have have deployed it on Windows. It wasn’t my Linux evangelism that was responsible for the decision it was pressure from both Oracle and our North American IT dept. which is running Oracle on AIX – They told them if you want support you have got to run on some kind of Unix. So for economy they wanted an x86 system, hence Linux.
So in away this decision took away from both MS and proprietary UNIX – Linux being the third option that was chosen.
I like many Linux users have a soft spot for Sun, the first Unix system I ever used was a Sun 3 workstation using SunOS as a grad student. I am also grateful for their contribution to open source: NFS, OpenOffice.org and the Gnome usability work. I wish them the best of luck but they are going to have to work out a consistent Linux strategy to prosper. Their emphasis on network computing gives them more of a chance than most people think.
My feeling that in the server market Linux is taking market share from both MS and proprietary UNIX. It us stopping they entry of MS into current UNIX shops on the back of Intel hardware and also beginnig to to chip away at MS servers in MS shops both at the bottom edge server end and at the mission critical top end (like my company).
If we are to to have real universal computer interoperability their are two alternatives – free open standards were many operating systems can interact (Linux, the BSD’s including Darwin, propietary UNIX and even BeOS) or a proprietary OS monoculture where a single monopoly i.e. MS decides on its on secret protocols which they then try to make their own proprietary IP. I know what I prefer
which is why I detest MS not because I think their software sucks but because it has now become a threat.
There is one tue operating system Unix, and Linux is its current manifestation – Amen (just a joke – maybe)
>Many consider Linux as the natural evolution over Unix.
There’s no reason to switch to Linux. Personally I think Sun Solaris is more advanced and mature than “toy-operating system” called Linux.
Linux was meant to be just a hobby, nothing serious and that’s the way it should be. Leave the hard and demanding work to the big boys.
Thank you, Torvalds. You’re not needed anymore. You’re done. You can now stop Linux development.
Personally I pred BSD-UNIX: Mac OS X, FreeBSD, OpenBSD.
Oh, man….
Is that troll flame-bait or what.
Anyhoo, I believe this article raises some very interesting points. We could debate this forever. However, only time will tell. My prediction is that Linux will take over for Unix and Windows will outlive Linux. Windows has power now and that has a lot to do with their expected success in the sever market. They have the resources to really put out a great product, and I think they will.
I love Linux and the ideal it bring to the OS community, but I use Windows out of nessecity. Except for my web server (which runs Red Hat) I use windows exclusively.
> Seeing old traditional Unices fading away with time, or at
> least losing their glorious role every day to Linux or
> Windows, truly saddens me.
I’ve never heard anyone say that before but you hit the fact right on the head. Oh how I wish DEC still existed!
>My prediction is that Linux will take over for Unix and >Windows will outlive Linux. Windows has power now and that >has a lot to do with their expected success in the sever >market. They have the resources to really put out a great >product, and I think they will.
No one in in their right mind thinks that that MS will beat Linux in the server market – not even Steve Ballmer. Which is one of the reasons MS is running scared at the moment.
Fair enough win2003 server is a good product. But it still stands no chance against Linux it cannot compete on price and support (yes support you can get better support with Google for Linux than with any fortune you spend with MS for “support”).
The latest performance comparisons paid for by MS are another Mindcraft scam. Not to mention the improvements that are in in the upcoming 2.6 kernel which will easily keep Linux ahead of Windows.
MS is doomed sell your shares now like Steve Ballmer
I always wonder why Linux developers want to be the server OS of the future. Why doesn’t the community realize that the fun stuff is a have a desktop OS which is fun to use, experiment with and make it into what people want to use.
Yet when you listen to the kernel group they seem hell bent on making Linux the OS for IBM mainframe use. Makes no sense. This just means no one will use it but the small niche of servers.
Make Linux the fun OS for desktop use. Suppose all the effort being put in MP Linux were to be put into actually making Linux easier to use, install and manage. Now then we would have something.
Linux’s rightful place is the desktop, not some dingy corporate server room.
bill
Interestingly, Sun is a corporate patron of the Free Software Foundation:
http://patron.fsf.org/2003-patrons.html
As are some other firms you may have heard of.
Do you really believe MS is gaining more market share? Don’t make me laugh. I see many people with WinNT 4.0 thinking about Linux that have never even heard of Win2k3.
Also, I see many people going from Windows to Linux while I have never seen anyone going to Windows.
Linux and the free BSDs are very similar – each have their own strengths and weaknesses and based on pure merit… I (my opinion) personally prefer FreeBSD, BSD developed alot of what UNIX is today. Sun and IBM furthered UNIX greatly, each in their own right. Both operating systems are nothing short of amazing (Solaris 9 and AIX 5.2.) Yet both vendors see the inevitable strength (irresistable force) of open – source software.
Two points if I may…
#1 Linux is not better, it is just free…
#2 CISC processors are now just CISC / RISC decoders that run really fast… and its inherent raw speed is overcoming its inefficiency. So Linux on Intel is cheap, cheap, cheap… that is why it is doing so well…
You think if Solaris x86 was free and open for development (under GPL) 5 years ago this would even be a debate, everyone would be using Solaris, same goes for AIX… But no one saw Linux coming and now no one can stop it because it has too much momentum, it is a buzz word like NT was 5 years ago (and XP is today.)
I think you will notice that RedHat benefits more than the others, they have started selling it for huge amounts of money. A sure sign for business idiots that it is “enterprise ready”. Which means RedHat sooner or later will become synonomous with Linux… like Xerox and copies (which I hope won’t happen, but it probably will.)
Solaris and AIX lack nothing, they are at the moment both superior to Linux (imo), but the Linux kernel is just improving over time and will surpass them, it is just a matter of time. The fact is binary compatability, price, and “buzz-word” power make this OS unstoppable. For better or for worse. Of course pure engineering genius and powerful OS’s will fall everytime to the current industry buzzwords… just how life is, its the class “prez” in the IT high school =)
Not to mention the improvements that are in in the upcoming 2.6 kernel which will easily keep Linux ahead of Windows.
I really hope so, MS server 2003 is a good heavy work and cluster computing operating system. The current Linux kernel can’t really hill itself like the 2003 server from MS. Not to mention MS server network balancing technology, it’s really paying off and will be looked for in future heavy computing solutions, I’m not talking about intranet/web servers but about scientific computing.
Why doesn’t the community realize that the fun stuff is a have a desktop OS which is fun to use, experiment with and make it into what people want to use.
Keep dreaming … while they release the next incomplete, unfinished, unpolished, buggy and inconsistent GNOME desktop. Huugh. (Better buy a QT License and develop an application on KDE).
a lot of excellent points here.
Linux, while not enjoying as long a past as other proprietary Unix/Unix-like OSs, has accelerated in development and has attracted a lot of young, intelligent developers.
I have seen (and implemented) Linux in a variety of environments, and all have stayed on Linux. While my resume can’t boast something like the success of Linux at google.com, I have introduced it to environments in:
-federal government
-universities
-web media (over 3 million unique hits/day)
-biotech (doing chemical reaction simulations)
buzzwords, hype, or not…it works very well in many situations, and I’m happy with it.
Maybe I’m not in my right mind, but….I think it’s just a matter of time. Win2003 is a very good product, and it’s plausible(sp?) that the next win server will be even better.
As far as support. Good point. That’s the thing…Microsoft will have to get more aggresive and I think that they can and def. will. They have proven to be very crafty and I think it’s just a matter of time.
I really hope so, MS server 2003 is a good heavy work and cluster computing operating system. The current Linux kernel can’t really hill itself like the 2003 server from MS. Not to mention MS server network balancing technology, it’s really paying off and will be looked for in future heavy computing solutions, I’m not talking about intranet/web servers but about scientific computing.
If you are talking about scientific clustering Linux is the only game in town. For a small to moderately large clusters Beowulf is the standard in scientific computing and is now dominating bioinformatics as well as physical science and engineering applications. For where SMP is required the latest SGI Linux machines using NUMA technology seem to be a very powerfull tool for non- cluster scientific computing. Then as for the most powerfull scientific computer in the world planned for manufacture – it is a Linux machine from IBM.
Windows is just not a player for scientific computing except for the most trivial applications.
I’ll admit we chemists (not on my home machine) tend to to use Windows a lot but when it it comes to serious quantum chemistry Linux or a proprietary UNIX is most commonly used.
AS for most physicists and mathematicians they won’t even have Windows on their desktops – nor even a word processor for that matter as they write their papers in LaTeX on Linux or Unix workstations.
Yesterday and today I played with the software Raid controls of SuSE Linux, and after this I understand you might be better off with a commercial Unix like Solaris.
I have SuSE 8.2. I needed more space for the home directory, but my partition layout did not permit this (physically /boot; /home; /; free space). So I moved all of /home onto /, then created a new partition in the free space (2.5 of 5 GB) and then used YaST to create a RAID array of the old /home and the new partition. This went all right, I copied all data to the new, 3.4 GB /home again.
But then I thought, well, why not add another partition of 2.5 GB, and while I was at it, remove the 2.5 gb one and replace it with a 5 GB one. So I started Yast, but it complained that it could not modify any existing RAID arrays.
So, I tried raidhotremove, but I did not manage to produce the right syntax and it does not have a manpage. Therefore I used raidreconfigure to remove the 2.5 gb partition (all data would still fit on the 900 MB one) but it failed and all data from /home was lost.
Apart from that I have most data on my Windows partition and OpenOffice now works again, this experience is not what I would expect from a system to be used in professional environments. To conclude, what would I expect:
1) Convert two regular partitions to a RAID volume (if possible without data loss)
2) Shrink and enlarge one of the RAID partitions
3) Add and remove Raid disks
4) Ideally this should all be possible while running, thus without any downtime, at least point 3 and I guess point 2 too, point 1 will be difficult.
Btw. I am talking about RAID 0 here, thus just concatenating two partitions together to one big one.
Probaly because many of these guys work for IBM, etc.
I personaly believe Linux is the best desktop OS atm for PCs. Others miss the apps or have problems managing them properly.
But……..If you are a gamer Linux leaves much to be desired. I have used winex but again it needs a lot of work.
Sun is a great company. It angers me when Linux dumbasses badmouth it. Just think of all the things they gave to the open source community. Where would linux be now without Java or OpenOffice.org for example? Let’s not forget that even XML was invented by Sun. Back in the midnineties they were there shoulder to shoulder with Netscape, together they had that sun/netscape server and they also had javascript, which now along with xul is at the heart of mozilla.
MS Windows makes the worst server of all major operating systems, however many use it. Why is this? It’s because people use Windows at home. Using it at home makes people more comfortable with the software, so they want to use it at work as well. They want to use it on the server and workstation because it’s what they know how to use, reguardless of whether it’s the best choice. It seems pretty obvious that many Linux users work in IT fields, and so these people bring Linux to work the same way Windows users bring Windows.
UNIX will continue to lose ground, not because it’s a worse OS, but because people feel more comfortable using Linux. Linux is what they know because they use it at home. Nobody uses AIX, HP-UX, or Unixware at home because they are too expensive for home use and usually run on expensive hardware. And possibly more important, there is less “home” software, such as games like Quake 3 and Return to Castle Wolfenstein, which is availble for Linux.
UNIX will probably continue to lose ground on the server, even if it is better, and this is why. Linux is popular on the server for the same reason Windows is–people are more comfortable with it because they know it from using it at home. UNIX is probably going to lose because of this, be it to Linux, Windows, or a combination of the two.
Charles,
You nailed it … that was exactly my point. Linux could be the best desktop os ever, if it were worked on from that perspective. If we build it the games will come, along with all the other apps.
But as long as Linux is steered to a server side solution, then it will, well, be delegated to the server niche. That spells doom.
I don’t argue that some of the 2,6 enhancements are good for the desktop, but the streering winds need to be changed. The real target should be the desktop.
1) It’s an open, standards-compliant OS. That means, APIs, “open interfaces”, which are oftem more important than just the source.
2) It’s a very powerful OS, Linux will need some 5 years -at least- to catch up.
3)It’s running on many computers, expecially on big computers which aren’t replaced that oftern.
4) Sun is committed to Solaris. Unlike, for example, HP, which doesn’t know it’s butt from it’s head.
Sun as a company still has a few arrows to spare. And the Sparc isn’t going anywhere, either: it’s an open architecture, unlike the Itanium. Ever wondered how did Fujitsu come up with a faster Sparc than the US III ? Because it’s an open design, where the SPARC group memebers can even contribute to it as a body, much like an OSS development.
man…I can see the Linux developers cringing right now! How hard does the 13 year old in Norway have to work for Linux to get credit for being a datacenter OS? “I’m sorry Mr./Mrs. director…I’m waiting on someone in the open source community to answer my posting to our mission critical problem. As soon as they answer I’ll be able to get that server back online”…..the only thing that kind of answer would get me is fired!
Say what you want to about binary compatibility..but when I can run ANY Solaris OS on a hardware that is 6 or 7 years old it gets some attention. I find it interesting when our windows department has problems running Oracle on certain HW….”I’m sorry department, if you want to run that version of Oracle you have to buy a new server type..not install a new OS…” whatever. I have a friend that just replaced 3 large SGI servers with 76 Linux server. Now they have power problems, cooling problems and space problems, not to mention they replace about 2 or 3 servers a week due to hardware failures.
Say what you will…..but that Sun server that was up for 1099 days (retired it last week) was a good example. It did it’s job, it just ran, doing it’s job. You want something to do..something to play with, install Linux and upgrade it every week. If you want a solid OS for your datacenter..install Solaris – it’s proven and it’s hardened.
Linux should be replacing Microsoft – not other UNIX’s. My 2 cents – they don’t mean anything…just my 2 cents.
tailgunner
your friend with cooling issues, power problems and hardware failures didn’t get those problems because of running Linux…he got those problems because he bought faulty hardware and didn’t do enough homework when designing it.
if he replaced 3 SGIs with 76 Linux boxes, then I assume he was doing some clustering. (yes ?) There are PLENTY of places who not only had no problems doing the same thing, but have seen major gains from going to Linux. at a MUCH larger scale.
btw, the caching dns/dhcp/mail machine I built in 2000 is still running at my old company. on a pentium 233.
if you want to say that Linux isn’t ready for the datacenter, then I’d suggest calling Google, Morgan Stanley, and Genentech and asking why they are doing it.
> Interestingly, Linux users seesm to love to badmouth Sun.
I’m a Linux user and don’t badmouth Sun. Comparing
HP-UX and Solaris, Solaris is *much* more Linux-like
.
And for the record, I hope SUN/Solaris does well. It is always nice to have choice. And we also have BSD for a choice :-)).
> Linux cannot do what Solaris can do (hot swap almost every
> component of a server for instance)
Please don’t mix apples and oranges. If it’s a solely OS feature to “hotswap” a CPU how come Solaris-x86 can’t do it ?!. Why can’t it hotswap a PCI card in a PC?
Hehe
. Sorry.. you are way off on this point. HotSwap must be supported first in hardware. And when (unexpensive) hotswap HW becomes popular I bet we’ll se a very efficient support for it in Linux.
Microsoft now has a license to stick the Windows program on top of a REAL operating system. Some may say this would be a good thing, but I have yet to see anything that Microsoft hasn’t bought, copied or stolen (Somewhere close to 99.999% of their products?) that has not suffered degradation due to Microsoft influence. I think Solaris will survive longer than some predict for one simple reason, SCO. I shredded my SCO media pack and licenses two days ago, my equivalent Solaris items are as intact as my Linux distros.