Dell CIO says Linux killing Unix: “Unix is dead,” said Randy Mott, CIO and senior vice president of Dell, in a keynote address on Thursday. Sun still sees Unix promise though. In the meantime, Linux targets the desktop as well as the server space. Reports here and here.
The Death of Unix, the Rise of Linux?
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
77 Comments
I am a bit disappointed with the speed of their progress, though. I am not thankless, just a bit perplexed at the pace freedos is progressing.
I have loads of old DOS software which I don’t want to part with. I even have an old copy of Ventura Publisher for DOS, from the time before Corel bought them. Shit, I even have a supported graphics card (Trident T-8900, ISA VGA card with 1 MB of RAM, a monster back in the day)!
Could be fun to see how these apps work with nowaday’s hardware.
Regarding Dell’s CIO:
I don’t put much weight into the rantings of a CIO who doesn’t sell either Linux or UNIX. If SCO or Sun were saying this, then I may think briefly about it before tossing into my mental bin of apathy.
Regarding the FreeBSD, OpenBSD, Linux, BeOS, Windows, etc. vs. each other discussion:
I have used Free, Open and Net BSD, and I have used Debian, Slackware, Gentoo, RedHat, SuSE, Mandrake and Caldera (good riddance to both as far as I’m concerned), and several other Linux distros, BeOS, and other operating systems.
I have not personally noticed much of a speed difference in any of them (well, except for Windows which runs fine at first, but quickly gets bogged down as you install applications). Benchmarks are fine, but if I can’t tell there’s a speed issue when using these OSes, then I couldn’t care less about the benchmarks (unless somebody is willing to pay me).
I don’t understand why some people have to hate Linux and love Windows, or love Linux and hate *BSD. They are all just tools; not religious sects.
I’ve said it a hundred times before, but when you receive a 150 piece tools set for Christmas, do you immediately throw away all the regular screw drivers in a fit of repugnance since phillips screw drivers are better? No, someday you may need to have the functionality provided by a regular screw driver. I view Windows, *BSD, Linux, etc. the same way. To view them with a theological zealotry, or worse, as you would a favorite football team, doesn’t seem very practical to me.
I have made money using Windows, Linux, Solaris, *BSD and some other OSes. Am I going to turn that money down because it came from Windows instead of Linux? Absolutely not. Am I going to knock platforms which are lucrative for me? Absolutely not.
I will never understand why people spend time writing for 1000000 flavors of UNIX when they can just write one version for Windows XP? They already have to support Windows XP anyway so it is effectively free to program a Windows version but writing all those 1000000000000 UNIX versions cost a ton of money. Not to mention the fact that if you write for Windows you can use .NET which allows you to write programs 10 times faster than for UNIX. Imagine how much simpler, cheaper, and better everything would be if everyone just ran Windows XP. No incompatibilities, cheaper software development costs, cheaper software, and so on. And Windows is a standard system unlike all those UNIX systems which all have proprietary quirks. Windows XP is also a lot more advanced than the UNIX versions and the user interface is a ton better. All the different UNIX versions including Linux are just a big waste of time and developer salaries.
This is well and good on a Linux system. Most of my comments here have to do with compiling software developed with GNU development tools under Solaris.
Reread my comment, I knew what you were talking about. What I saying was expect the problems on Solaris to get worse not better.
I’m not sure if you saw my comments about OS X before or what, however in this case I am talking about Solaris.
I was talking about getting it to compile for BSD. I.E. OSX is BSD based and Fink maintains a good list of Linux to BSD patches.
__________
As far as the top500 Altix isn’t on the list since it isn’t deployed yet. It came out for sale like a week ago; and demod only about 4 months ago. BTW a Linux machine (much slower hardware) is at #5 on the Top500 list, and another at #8. I was actually suprised to see Dell checking in at #22 with a Xeon box running Linux. Sun first checks in at #145 running just about 12x slower than the Linux box at #5. I don’t think the top500 proves your main point at all.
Can’t realy dispute pretty much everything Bascule is saying with regards to Linux’ technical inferiority (especially on x86) compared to established commercial Unices. Even Dennis Ritchie (everyone knows who he is, right?) in a 1999 issue of IEEE Computer magazine said basically that he looked at the Linux source code and thought that a lot of it was crap and he didn’t think that Linux deserved to be particularly successful in the future (I will hunt down that interview and post excerpts here one of these days, probably – he slams Linux really hard.). Of course, some of this could have been sour grapes that his own Inferno and Plan 9 projects had not taken off (he should have open-sourced them) and this was 4 years ago now too.
When MSFT is introduced into a discussion here, you often hear someone say “the best technology doesn’t always win” when referring to Windows’ domination of its market. Well, now the same principle is applying itself in the Unix world – Linux has so much hype and mindshare behind it that it is becoming a de facto standard, even though it probably does not deserve to be based on its technical merits alone. Really sucks to be on the receiving end of that principle from a fellow member of your own Unix “family”, doesn’t it?
The only hope for the hardcore ‘real’-Unix community (e.g. Bascule) is that with such a huge amount of money and manpower going into Linux right now (say, compared to the *BSDs), it will improve by leaps and bounds and eventually become a lot better technology than it is now so as not to drive them completely up the wall when/if they have to work with it.
The people who have actually used completely dumb terminals / X terminals will tell you there were huge advantages to everything local. One of the biggest is the machines are essentially disposal. There is 0 connection between the machine you are on and any information.
The big advantage for companies in this configuration is a) the savings for help desk… costs
b) the savings on real estate costs
(a) has been beat to death and everyone understands it. (b) is what Sun is now pushing, and implementing internally. If you use dumb terminals and make a commitment to not have anything on paper then people don’t need their own desks. If you consider:
a) people working different hours
b) some people being sick, vacation, transitioning between departnments and or companies…
you end with most desks idle most of the time. You might very well be able to reduce the number of desks by up 50% by just not making them personal desks.
A 50% real estate reduction expense for office workers would be immense saving dwarfing by a long shot anything having to do with price differences.
More importantly once people genuinely aren’t keeping anythign at work, telecommuting and working remotely become more plausable.
Secondly, with true client server you can achieve a much higher utilization rate on your corporate CPU power and diskspace. Even if you have to pay 2-3x as much for the disk space or CPU power it still is a savings.
The problem with Citrix is that the machines on the other side and truly dumb machines. They have moving parts, they can meaningfully break, they need to be maintained by a help desk…. You don’t get nearly the labor savings nor the huge increase in reliability. Now compound that with the high cost per seat for the applications and I don’t see Microsoft’s client / server setup as offering nearly the value of Sun’s.
As for the “smart session” cards, the purpose of those is to authenticate a little better than passwords. They aren’t needed for the solution; they are an optional feature not a bug.
I’d really suggest those people interested in thin clients take a look at the materials on Sun’s website.
Yes, Gil Bates! This ought to be the first (maybe last) time that I agree with you 100%. And yes, it sucks, but no, UNIX won’t die. If I got a dime for every time someone said that *BSD is dying, NetWare is dying or UNIX is dying (actually, it was Dell manager, maybe Michael Dell himself that once said that NT will kill UNIX) I would be a millionaire now.
Even Dennis Ritchie (everyone knows who he is, right?) in a 1999 issue of IEEE Computer magazine said basically that he looked at the Linux source code and thought that a lot of it was crap and he didn’t think that Linux deserved to be particularly successful in the future
It was Ken Thompson. This is what he said:
Thompson: I view Linux as something that’s not Microsoft—a backlash against Microsoft, no more and no less. I don’t think it will be very successful in the long run. I’ve looked at the source and there are pieces that are good and pieces that are not. A whole bunch of random people have contributed to this source, and the quality varies drastically.
My experience and some of my friends’ experience is that Linux is quite unreliable. Microsoft is really unreliable but Linux is worse. In a non-PC environment, it just won’t hold up. If you’re using it on a single box, that’s one thing. But if you want to use Linux in firewalls, gateways, embedded systems, and so on, it has a long way to go.
I think this says a lot more about Ken Thompson than of Linux.
I will never understand why people spend time writing for 1000000 flavors of UNIX when they can just write one version for Windows XP?
Windows 2000 is a pretty good OS, but XP absolutely sucks. One answer to your question however is power. XP doesn’t have much by comparison. Control is another good answer.
Not to mention the fact that if you write for Windows you can use .NET which allows you to write programs 10 times faster than for UNIX.
Not really. It depends on what you are writing of course. For example, a text parser is much easier and faster to write in languages like Perl or Python than they are in .NET. Always pick the best tool for the job instead of continually using a favorite.
Imagine how much simpler, cheaper, and better everything would be if everyone just ran Windows XP.
And imagine how much less stable the internet would be.
No incompatibilities, cheaper software development costs, cheaper software, and so on.
I’ve written software for both UNIX and Windows, .NET makes Windows programming easier that is has been in the past, but development costs are not cheaper. By the way, Windows is full of incompatibilities. I’m dealing with two driver incompatibilities at work right now.
And Windows is a standard system unlike all those UNIX systems which all have proprietary quirks.
It’s not standard, rather proprietary.
Windows XP is also a lot more advanced than the UNIX versions
No it is not. Windows is probably the least advanced OS I’ve ever used (I haven’t used SkyOS yet).
and the user interface is a ton better.
…and a ton slower. I think OS X has the best interface (talking from a technical standpoint; not an aesthetics one). X and Windows GDI both suffer from some of the same problems. However, X is much more robust and flexible than Windows’ interface is, so I think it is better from a technical standpoint as well.
All the different UNIX versions including Linux are just a big waste of time and developer salaries.
Your opinion is probably formed from the fact that you don’t use UNIX.
How does a Wintel slave suddenly feel like talking about Linux (I am specifically not mentioning GNU/Linux since the CIO of Dell used ‘Linux’).
Btw: I know that Dell has been toying around with Linux for a while.
Speaking of Dennis Ritchie, I came across http://cm.bell-labs.com/cm/cs/who/dmr/otherunix.html a few days ago.
Now that sets me wondeting: which Unix was the Dell CIO referring to:
The fire extinguisher, the TV antenna, or……
As far as the top500 Altix isn’t on the list since it isn’t deployed yet. It came out for sale like a week ago; and demod only about 4 months ago.
I was merely pointing out that it is far from the world’s most powerful computer… it isn’t even the most powerful system SGI manufactures…
I don’t think the top500 proves your main point at all.
That’s because you’re missing it entirely. Sun makes *servers* which handle I/O bound tasks (i.e. moving a lot of data around without doing a great deal of processing on it) While this is great for… a central codebase repository, a central mail server, etc. most modelling tasks aren’t particularly I/O bound because the majority of time spent in modelling isn’t spent moving data around, but processing it.
As I mentioned before, a Sun Fire 15k has a peak system bandwidth of 172.8 GB/s, and a sustained bandwidth of 43.2 GB/s. Its I/O bandwidth is 21.6 GB/s sustained. This is certainly overkill for most modelling applictaions. Where I work, our modelling software consumes about 20-30Mbps, whereas Sun is offering ~173,000Mbps in the Sun Fire 15k.
I found it interesting reading everyone’s responses. I think that when it comes to Unix vs. Linux there are somethings that people have to keep in mind.
First, Linux was created and runs well on X86 hardware.. no question there. However, once you pass 2 processors, it does not really scale linearly. There is no way that it could scale to 64 or even 256 processors, the way Solaris, Tru64, or IRIX can. And while you may be able to get similar levels of performance out of hundreds of PC’s linked together.. does it make a lot of sense? A lot of companies will have to learn the lesson all over again that a few big machines are easier to manage than hundreds or thousands of smaller machines. In this economy, more systems to manage is not a good thing. Companies today want to do more business with less (fewer systems, fewer people, and less space). A few people brought up the Sun Ray’s. They are an excellent way of reducing the number of resources required to do business. Sun has forced the vast majority of it’s own employee’s onto Sun Ray’s. And they work! I know, because I use to work at Sun. Why spend money on PC’s every year (whether they run Linux or something else), and the management services needed to support them when you can run and control eveyrthing from a few servers. It’s a great solution. While it does have an upfront initial cost, it’s nothing compaired to the endless number of PC’s and MS licenses companies buy into over time.
Second, Linux has recieved a lot of attention over the years and it has helped it gain mindshare. However, mindshare does not equal market share. While there may be more people installing Linux out there, it does not mean that it’s being used to benefit the corporate world in the sense of replacing Unix. Linux has it’s place on the desktop and for taking care of low profile tasks (file,print,NFS, etc.).
Third, many Linux users have not worked with or used a commercial Unix for an extended period of time. Many Linux users talk about how much faster Linux is compaired to Solaris. Now, if you ask those users what Sun box they are compairing their PC to, it’s usually a Sun workstation from 5 or more years ago. Rarely are they compairing fairly. Just about anything new will be able to outperform an old workstation from over 5 years ago. DUH! A Sun SparcStation 20 can beat the pants off of a 386. But if you were to compair a new PC and a new Sun workstation like the Sun Blade 2000, you would see that Solaris (especially Solaris 9) is not slow. Now if you compair a new Sun server to a PC server, you’d see an even huger difference. Sun hardware is very reliable and scalable.
Third, Linux has a lot to learn about compatability. When you install Linux, there is a wide variety of software included that many Linux users refer to when compairing against commercial Unices. You can recompile the vast majority of free software on any Unix or Unix-like OS. The problem arises in the fact that a lot of software written on Linux depends on Linux specific behaviors or compiler options. And that’s not a good thing. Good Unix software should recompile on any Unix with little or no modifications. And this is where the commercial Unix’s are way ahead of Linux. Take Solaris for example, you can run applications written from Solaris 2.6 and still run them on Solaris 9 without a recompile. You can run 32bit apps in 64bit Solaris, no problem. The point is software compatability. If Linux wants to really take on the commercial Unix market it has got to work towards compatability. Even Apple was forced to realize this with Mac OS X, that’s why they now have a beta X11 server. And don’t talk to me about commercial Unices not including free software.. that’s BS. Solaris has included a CD specificly for GNU software, GNOME, KDE, etc… all compiled with the Forte compilers in Solaris 9. Compaq/HP has a freeware CD that comes with Tru64. And many other commercial Unices have done the same.
Forth, fancy software does not equal better OS. By that I mean that just because most Linux distros include lots of software, does not mean that Linux itself is better than any other Unix or Unix-like OS. When you look at the functionality of the Linux kernel and compair it to the major commercial Unices, it’s a good 10 years behind. Linux lacks a good VMM, threads implemenation, SMP support, Dynamic reconfiguration, etc. I mean, you still have to recompile the kernel and/or modules. For Solaris, you just do a boot reconfigure. And that’s another thing.. Solaris will detect new devices and automaticly load the support for them. Have any of you looked at Sun’s Sun Fire line? You can add processors and memory on-line, no reboots. You can even put a board in with more CPU power and memory.. move your system image over, and then remove your old CPU/memory board.. without a reboot! You can’t do that with Linux.
In the end, Linux has a long way to go when it comes to taking on the Midrange to High-End server market that commercial Unices dominate. And there may be more people buy/installing Linux out there.. but that’s because the bigger Unices have a long life-cycle. I mean there are more people out there running Solaris 2.6 on old sparcstation equipment than you can shake a stick at. And that’s really Sun’s problem. Their hardware and software just works too well;)
If Linux people really want to see MS fail, they need to embrace their bigger Unix cousins. Linux should work harder towards desktop dominance than to try and compete with bigger Unices, because they are too different things. If you want people to buy and love the Penguin, then focus on that end. Then you can worry about the other things. But trying to do both at the same time is a difficult and uncertain path. Look at Apple, it has the user experience first in mind.. and while it would be cool to see them go into the server market (both low-end and high-end), they don’t have the hardware or the extensions to its OS to make it work.
Personally, I think that Apple has a better chance of dominating the desktop market than Linux does. Simply because it is focused on making a product that is aimed at the consumer market. Linux is not focused on any particular market place, and that will always be it’s problem.
(Soap box mode off)
UNIX is not dead. We rely heavily on Solaris where i work. And i know of very many companies that rely on commercial UNIX offerings.
However if Linux continues to become stable and mature and stays at a fraction of the cost of commercial offerings then it will make further inroads.
As long as the large vendors such as RedHat dont start getting greedy and go down the Microsoft path. Although i guess as long as the stuff stays under the GPL then people have a choice.
“Unix is dead,” said Randy Mott, CIO and senior vice president of Dell.
A spokesman from Dell, a company that manufactures solely x86 equipment, is saying that “Unix”, which I think is being used here as an umbrella term for all Unices which run on proprietary RISC architectures, is dead. I’m so shocked I don’t know what to say.
“The idea that Linux obliterates Unix is dead wrong,” said Jonathan Schwartz, vice president of software at Sun.
Linux has already won hands down in one department: mindshare. I’m so sad to see that blind and rabid zealotry has managed to win Linux enough mindshare to make it the de facto standard. This is a two way street too, not only were these zealots spreading their blind love of Linux, they were also lumping a great deal of undue criticism upon other operating systems. Solaris is slow (in fact let’s call it Slowaris)… BSD is slow… sure, but have you ever used them? No?
I find it amazing how the crappy little OS that could has managed to attract so many big names. SGI and IBM have written elaborate abstraction layers to graft their filesystems onto the Linux VFS, when the FreeBSD VFS would’ve made a much better candidate as far as similarity to the originating operating systems go. I don’t see how this can be due to licensing issues… the GPL protects SGI’s and IBM’s development investment for those filesystems, but nothing prevents them from releasing the filesystems under the GPL then linking them with BSD licensed code. Linux just had greater mindshare, thak you zealots, and because of that Linux gets the toys.
It’s certainly true that Linux does not obliterate Unix. Solaris has Linux beat hands down in terms of raw throughput, as does Sun hardware when compared to its x86 counterparts. Sun has been the king of I/O for several years and I don’t see that changing any time soon.
Well, while the Dell CEO is proclaiming Unix is dead, my Solaris servers will be reliably humming in the background, delivering mail and serving files. At least where I come from, it’s not dead yet…
Well, while the Dell CEO is proclaiming Unix is dead…
Excuse me, I meant CIO…
Blah blah blah. I’m sick of hearing about Linux on the desktop. It’s the Emperor’s New Clothes of desktop operating systems. It needen’t be, just in case you think I’m a natural born Linux nay sayer but it is.
Unix seems to split apart and converge constantly. Linux is just another incarnation of this process. And yes, Linux is Unix – a horribly basterdized and mutated Unix, but Unix never the less.
so linux is going to take over the back room, and the guys and gals in IT aren’t going to think, eventually, “why don’t we just run this on the desktop too?”
I find it amazing how the crappy little OS that could has managed to attract so many big names.
I don’t. IBM needs Linux to keep it’s high profit ‘accounts’ from moving to Microsoft solutions.
This is what their ‘accounts’ made IBM understood: No total package from IBM, No need for IBM.
so linux is going to take over the back room, and the guys and gals in IT aren’t going to think, eventually, “why don’t we just run this on the desktop too?”
Not after they see it for themselves.
When all is said and done, those high accounts really don’t have a clue about computers. Nor do they really care oh so much about cost – if they think its worth it. Linux is new, its trendy, and the most spoken OS buzzword these days. It doesn’t matter if its locking isn’t as fine as Solaris, or if its not as secure as OpenBSD, or if it has a great filesystem like FreeBSD’s with SoftUpdates. All that matters is that there are enough empty cans rattling as loud as possible, getting linux into the news and into “business people’s” minds as the next good-thing?.
Its taken it this long to catch up with commercial unicies, and even now the only really good thing it has going for it is that its free. Which when you consider the BSDs the free thing doesn’t really seem so unique anymore. Pretty much when I look at it, linux is a poopy OS that really doesn’t stand out from the other Unicies execpt for the fact that their community is just plain obnoxious.
Calling GNU/Linux a crappy little OS is no were near the truth. I understand what your trying to say by rabid zealotry but the fact remains that the one thing that sets linux appart from all other open source NIX’s is exactly that zealot userbase. The zealots are the ones making sure the system is not just secure but also free. Both Solaris and FreeBSD ARE very slow on the x86 platform. As far as ‘NIX’s go, Linux definitly more advanced than FreeBSD, and yes I have not just used but administered a FreeBSD server. Solaris on the other hand runes well on a SPARK but in this day and age of falling tech budgets, who will pay for a SPARK when you can get a deacent x86 server running Linux for less than half the price?
Maybe you should actually look at the features that come STANDARD on the SPARC, PA-RISC and RS/6000 machines instead of spouting zealot “everything that isn’t x86 sucks” rants.
All that is happening now is that these ultra high end UNIX servers are replacing Mainframes, Mainframes are now a small niche and as x86 becomes more reliable the were be a gradual move to them.
John Galt: Linux is Unix
Gee, I hate when people say that. Linux is not UNIX; Linux is Linux.
CypherPunk: Both Solaris and FreeBSD ARE very slow on the x86 platform……yes I have not just used but administered a FreeBSD server.
Please, clean your lies.. FreeBSD is not very slow, which it’s very fast.
Well, notice that the only 2 workstations that sun offers are the 150 and the 2000. There is no intermediate machine that can run or even compete with a Wintel machine running Linux or FreeBSD.
Linux is great because it’s GPL. That’s all. BSD style licenses have helped Microsoft take the entire BSD TCP/IP stack and put it into Windows.
GPL is the most empowering license ever devised and that is why Linux will succeed where others have failed. Thank God that RMS had the foresight to write software released under the GPL.
Solaris on the other hand runes well on a SPARK but in this day and age of falling tech budgets, who will pay for a SPARK when you can get a deacent x86 server running Linux for less than half the price?
Someone who needs the support offered by Sun, wants a unified OS and hardware setup (a la Mac) and is willing to pay for what they want.
Both Solaris and FreeBSD ARE very slow on the x86 platform.
This is in fact your opinion, and its very disingenious to state it as a fact. I have a couple of distros of Linux, and in my experience, even 5.0-current (with debugging feautures compiled in) has been noticably faster than any distribution of Linux I have tried. Stop the presses! FreeBSD is faster than Linux? No. Of course it’s just my personal experience, and I’m not stupid enough to believe everyone has the same computer, same set-up and will have the same results. I can understand that after using a computer for a while, you may start to have a social disconnect (or a reality disconnect), but everything that applies to you, does not neccesarily apply to everyone else.
Not sure if you are trying to make a joke or not, but it is spelled SPARC.
First of all, do you have anything to backup what you’re saying about FreeBSD being slow on the x86 platform, because I believe this is false. Second of all, I believe it’s called SPARC, not SPARK.
Solaris on the other hand runes well on a SPARK but in this day and age of falling tech budgets, who will pay for a SPARK when you can get a deacent x86 server running Linux for less than half the price?
Someone who needs the support offered by Sun, wants a unified OS and hardware setup (a la Mac) and is willing to pay for what they want.
Both Solaris and FreeBSD ARE very slow on the x86 platform.
This is in fact your opinion, and its very disingenious to state it as a fact. I have a couple of distros of Linux, and in my experience, even 5.0-current (with debugging feautures compiled in) has been noticably faster than any distribution of Linux I have tried. Stop the presses! FreeBSD is faster than Linux? No. Of course it’s just my personal experience, and I’m not stupid enough to believe everyone has the same computer, same set-up and will have the same results. I can understand that after using a computer for a while, you may start to have a social disconnect (or a reality disconnect), but everything that applies to you, does not neccesarily apply to everyone else.
Not sure if you are trying to make a joke or not, but it is spelled SPARC.
(hoping this isn’t a double post, waited 5 minutes…)
Your right it is SPARC 😮 But I have found that running FreeBSD on an x86 is slower than Linux. I’m not saying this cause I’m a Linux nut, I do have my reasons for this. When I first started working for Saios here in Luxembour we were running our Intrenet server on FreeBSD. The then sysadmin decieded to ‘try’ and upgrade to a more recent version and screwed up. I sujested Mandrake for It’s great extranet system and low and behold, on the same hardware we were able to run PHP scripts at least 20% faster. Both systems were running apache by the way.
Calling GNU/Linux a crappy little OS is no were near the truth.
GNU/Linux? I see RMS has brainwashed you…
I understand what your trying to say by rabid zealotry but the fact remains that the one thing that sets linux appart from all other open source NIX’s is exactly that zealot userbase. The zealots are the ones making sure the system is not just secure but also free.
I find the base of zealots consists primarily of recent converts, most of which have never used any operating systems other than DOS and Windows.
Both Solaris and FreeBSD ARE very slow on the x86 platform.
I’m getting closer to releasing this code, but I’d prefer not to yet. However, here are my benchmarks of process management and heap performance in Linux 2.4.20 w glibc 2.3.1 versus FreeBSD 5.0-RELEASE, on a dual 1.53GHz Athlon system with 512MB RAM:
1 process:
Linear allocation/deallocation speeds (262144 * 1024-byte blocks)
FreeBSD: 0.628473 seconds / 0.144241 seconds
Linux: 0.842112 seconds / 0.125834 seconds
Pseudorandom allocation/deallocation speeds (262144 * 1024-byte blocks) set 1:
FreeBSD: 0.891263 seconds / 0.136810 seconds
Linux: 1.1062513 seconds / 0.108821 seconds
Pseudorandom allocation/deallocation speeds (262144 * 1024-byte blocks) set 2:
FreeBSD: 0.761018 / 0.092014 seconds
Linux: 1.261043 / 0.156823 seconds
Linear small blocks test: (1024 * 64-byte blocks, total from 1024 iterations)
FreeBSD: 0.561032 / 0.082136 seconds
Linux: 0.662042 / 0.071093 seconds
Pseudorandom small blocks test: (1024 * 64-byte blocks, total from 1024 iterations)
FreeBSD: 0.618093 / 0.101026 seconds
Linux: 0.819057 / 0.128194 seconds
Time required to spawn 1024 processes:
FreeBSD: 0.198692 seconds
Linux: 0.168305 seconds
1024 processes:
Linear allocation/deallocation speeds (1024 * 1024-byte blocks * 256 processes)
FreeBSD: 0.863209 seconds / 0.241850 seconds
Linux: 2.793860 seconds / 0.231801 seconds
Pseudorandom allocation/deallocation speeds (1024 * 1024-byte blocks * 256 processes):
FreeBSD: 0.901403 seconds / 0.231809 seconds
Linux: 2.680194 seconds / 0.210360 seconds
I don’t think these numbers should be too surprising. The deallocation numbers in the multiprocess tests also include the teardown time for all the processes spawned, so those aren’t a pure reflection of VMM/heap performance. Everything else should be. The conclusion you can draw from these numbers is this: FreeBSD has the better VMM, and Linux is better at creating new processes quickly. In tests which result in a large number of simultaneous segment size changes, the Linux VMM falls flat on its face.
As far as ‘NIX’s go, Linux definitly more advanced than FreeBSD
The VMM certainly isn’t. FreeBSD also has a number of “advanced” features such as stateful I/O multiplexing and event handling as well as filesystem metadata caching.
It’s clear in many areas (primarily process management) that Linux is winning, but could you say the same thing 5 years ago? The reason for this is, of course, mindshare, which comes from its overzealous userbase.
and yes I have not just used but administered a FreeBSD server.
Why do I have no doubt in my mind that my Linux experience is several orders of magnitude more than your FreeBSD experience?
Solaris on the other hand runes well on a SPARK
Perhaps you mean SPARC.
but in this day and age of falling tech budgets, who will pay for a SPARK when you can get a deacent x86 server running Linux for less than half the price?
Someone with a large resource (such as a codebase) which isn’t easily distributed amoung multiple servers. Where I work, we use our Blade 1000 and 2000 workstations for mesoscale atmospheric modelling.
Sun will also continue to dominate in any areas where reliability and throughput are key, such as high volume mail servers.
Well, notice that the only 2 workstations that sun offers are the 150 and the 2000. There is no intermediate machine that can run or even compete with a Wintel machine running Linux or FreeBSD.
Meaning Sun has no middle market? Hell, even the Blade 100/150 were outsourced. The only way Sun even remotely cares about the low end is deploying large thin client networks. Sun’s old slogan was “The network is the computer” Their approach was to deploy a large number of thin clients (Sunrays) which would run off a single high powered central server. This greatly diminishes the TCO of a large network because there is only a single server which needs to be maintained, so one admin can easily maintain a network of several hundred thin client workstations, whereas a similar setup with seperately installed and configured systems may require a dozen administrators.
Unfortunately, Sun has had a great deal of trouble selling Sunray setups, due to the obvious high initial cost.
Linux is great because it’s GPL. That’s all.
Yes, the license makes the software, regardless of how crappy the implementation is. Your logic is infalliable.
BSD style licenses have helped Microsoft take the entire BSD TCP/IP stack and put it into Windows.
Is this a bad thing? Personally I’d like Microsoft to do whatever they can to improve the performance and stability of Windows.
GPL is the most empowering license ever devised and that is why Linux will succeed where others have failed. Thank God that RMS had the foresight to write software released under the GPL.
Oh please, what did RMS ever write besides the most bloated, buggy, and useless editor on the planet?
I’m not sure if you are trolling or actually posting.
Linux has already won hands down in one department: mindshare. I’m so sad to see that blind and rabid zealotry has managed to win Linux enough mindshare to make it the de facto standard.
No, a coordinated effort to become the defacto standard. In particular examining the failures of their own and of other systems objectively and looking for opportunity.
This is a two way street too, not only were these zealots spreading their blind love of Linux, they were also lumping a great deal of undue criticism upon other operating systems. Solaris is slow (in fact let’s call it Slowaris)… BSD is slow… sure, but have you ever used them? No?
Well I’m not super experienced with BSDs; though I’ve used a 1/2 dozen or so currently use Darwin and keep up with them. As for Solaris there I have a solid decade of almost daily use. I don’t think Solaris is slow nor BSD is slow. OTOH Intel has done an amazing job of pushing their CPU performance faster than everybody else in the industry. Combine that with the overerly long lifespan of Sun boxes and often the user experience will seem quite slow.
I find it amazing how the crappy little OS that could has managed to attract so many big names. SGI and IBM have written elaborate abstraction layers to graft their filesystems onto the Linux VFS, when the FreeBSD VFS would’ve made a much better candidate as far as similarity to the originating operating systems go. I don’t see how this can be due to licensing issues… the GPL protects SGI’s and IBM’s development investment for those filesystems, but nothing prevents them from releasing the filesystems under the GPL then linking them with BSD licensed code.
A very big thing does. None of the BSDs would accept the patch into the main kernel source if it were GPLed. Which means they would have to patch again and again and again…. There is another big thing that stops them. Linux is SystemV based which has always made porting Solaris -> Linux easier than Solaris -> BSD.
Linux just had greater mindshare, thak you zealots, and because of that Linux gets the toys.
Also Linux cared about the toys more. The companies that showed interest in Linux even when they produced minor toys got great press. But yeah Linux won the early mindshare battles got the toys and from there….
It’s certainly true that Linux does not obliterate Unix. Solaris has Linux beat hands down in terms of raw throughput,
Actually no, sorry ain’t true anymore. The <href=”http://www.sgi.com/servers/altix/“>fastest computer in the world is a Linux box. 64 Itanium2 processors per node, each processor supporting up to 8 gigs of very high speed ram; disk control subsystems allowing reads at 2gBytes / second. Sun doesn’t have anything in that speed range.
as does Sun hardware when compared to its x86 counterparts.
Well yeah. x86 doesn’t compare to Sun hardware, so what. Linux on x86 beats Solaris on x86. We’ll have to see how Linux on Sparc III does in speed tests. I’d hope Solaris could beat Linux on their home platform or they are really toast.
Sun has been the king of I/O for several years and I don’t see that changing any time soon.
Sun has never been the king of I/O. SGI has always had them beat. Sun has always good I/O, but that does not a king make. What Sun has always offered was a balanced package at an attractive price.
I’m not sure that benchmark is all that important. The entire VM system is replaced in 2.5. I’m not talking an upgrade here this is a “rip out the old code, and rewrite from scratch” type upgrade.
As for GNU/Linux; Linux as well as most Linux software has been released under the “GNU Public License”. I don’t think its too much a stretch to argue that Linux is part of the GNU project. Especially given the fact that Linus himself confirms that he was inspired by RMS to release under the GPL and did consider his work to be in line with the project as a temporary kernel until Hurd was finished.
first off:
bsdrocks (IP: —.ks.ok.cox.net) I am not an x86 nutcase. I mearly pointed out that it’s a cheaper architecture than SPARC. Personaly I’d love a SPARC but can’t afford one.
Secondly:
Bascule (IP: —.client.attbi.com) I have no doubt that your experiance in Linux out does mine and have never stated otherwise. As for your snide remarks about my spelling of SPARC and being brainwashed, is that the best you can do? Your first post reminds me of exactly those zealots you seem to hate so much. Sun servers are without a doubt excelent machines. I myself have only had the chance to work with an UltraSPARC 10 and already found myself inpressed. But what I pointed out hardly means that I agree with Randy Mott about Linux killing UNIX.
I have been using RedHat Linux since ’99 as my main OS and what have I really sacrificed?
Document compatibility?
The first thing I did after I got my RH5.1 system online was to download StarOffice 5.1. I used it until OpenOffice hit 1.0 and I never had a problem exchanging Word, Excel or PowerPoint files with any of my classwork and business contacts. Now I have OO 1.1 that installs with the latest RedHat and displays text using a FreeType engine that is on par with what I’ve seen of OSX and shows why Windows has its so called ‘ClearType’ disabled for small fonts
Multimedia display?
Realplayer, Flash, MPEG, MP3 and others all since almost day one. Now with the work that has gone into AVIFILE and Xine even many of the WMP formats too.
Easy software install?
I’ll admit that I have spent a number of evenings hunting around rpmfind.net to meet the dependencies of something I wanted to install. But face it – anything that is not a pre-alpha or a weekend project is going to have a well packaged RPM that will install on any recent RH distro with minimal dependency hunting. Now: most of the packages I need on a daily basis are now in the apt-rpm archives on a number of apt-rpm servers. Touch that with 50 click-next-to-continue dialog boxes.
Hardware detection?
Kudzu detects just about anything that has a Linux driver at the first boot after installing it. No reboots required. I did have some early problems with my parallel port ZIP100 and CD-RW back in the day. Sadly I still have to be choosy about hardware but that does help me avoid cheep crap. And thank heaven for nVidia
Graphics, video and 3D modeling?
Unix is not dead of course, since Linux is also a Unix variant, but since by Unix they refer to Unixes from Sun, IBM and so on, that’s pretty much true. That will die.
Intel is coming with processors for high-end machines. You can easily port Linux for that. Once Linux becomes better other solutions will become irrelavent.
Check out the latest profit figures from Sun. Sun is going down. That’s why they are suing Microsoft, but they will not get anything, since the reason why Sun goes down is not Microsoft, but actually linux and open source.
Blender was one of the first Linux programs I downloaded and some people have done astounding things with it. GIMP may not be PhotoShop but it is certainly on par with PaintShopPro and light years beyond the crud that comes with most scanners. I have only player with them but Cinnerella and Kino look to be very capable video capture and editing programs. My music collection was built using GRIP a program with one of the most streamlined UIs ever.
Games?
I’ll admit to keeping a Windows install for that one But I also have an even larger library of independent games for Linux. I do waste a lot of time even in Linux.
UI and ease of use?
This is subjective. But I get a sick feeling in the pit of my stomach when confronted with the inflexibility and ugliness of that other OS . . . No virtual desktops, No sensible copy paste, No edge snapping, very few usefully application installed by default.
This is no fantasy or marketing gimmick but my personal experience – warts and all. Can Linux be a great Work station and general purpose platform? I think I can confidentially confirm that. Can it be frustrating? It has been in the past. Is their room for improvement? Well DUH is anything perfect? Are their general ease of use things that are lacking in comparison to Windows and Mackintosh? Yes I wish that the device driver ABI was stable or that drivers could be moved to User space to make installing drivers less of a pain. Also I would like to see more commercial applications targeting Linux – but that is defiantly starting to happen. Are their areas where Linux is ahead of Windows as far as basic usability?
Drum roll please . . .
Yes. Windows users I know are almost always blown away by APT and the Synaptic package management GUI. Simply using multiple desktops by default puts Linux on a new level when it comes to running multiple complex programs at once.
So in conclusion I must say that their is most definitely computing life outside of Apple and Micro$oft. Linux is improving as a platform – very rapidly. Will Linux become a serious player on end user systems? In time – its still a long road but its being traveled.
I’m not sure that benchmark is all that important. The entire VM system is replaced in 2.5. I’m not talking an upgrade here this is a “rip out the old code, and rewrite from scratch” type upgrade.
The VMM is only half of the equation… there’s also the heap implementation to worry about. 2.3 includes a newly revamped heap, and further proves that newer is not always better… improvements over glibc 2.2 were minimal. FreeBSD’s heap implementation integrates tightly with its VMM implementation… they were written by the same people, after all. Two completely different groups develop glibc and the Linux kernel, so such integration is pretty much impossible. In fact, the opposite tends to occur, see http://www.cs.helsinki.fi/linux/linux-kernel/2001-31/0264.html
I’ll benchmark 2.6 when it is released.
As for your snide remarks about my spelling of SPARC and being brainwashed, is that the best you can do?
How are my remarks about the spelling of SPARC “snide”? I corrected your spelling, as did several other people in this thread…
I apologize if you take personal offense towards my disdain for the phrase “GNU/Linux” I see no need to recognize the “work” of GNU and RMS, when all it has done is made my life harder. Most of my systems have two completely disparate toolsets installed: tar and gtar, make and gmake, lex and flex, sh/ksh and bash, simply because GNU chooses to modify tools to the point that it breaks intercompatibility, rather than reinventing from scratch. At least with the latter I wouldn’t need to edit shell scripts which assume /bin/sh is bash, go through C code and remove C++ style comments because they assume cc is gcc, or replace all occurances of lex in a Makefile with flex because the author assumes lex is flex.
Your first post reminds me of exactly those zealots you seem to hate so much.
I do harbor a great deal of scorn, vested primarily upon GNU for the reasons I’ve just stated. You can call me “zealous” in my disdain for GNU, but my zeal stems from experience, not from ignorance.
GIMP may not be PhotoShop but it is certainly on par with PaintShopPro and light years beyond the crud that comes with most scanners.
Finally, a realistic assessment of the Gimp’s abilities.
I do not take offence towards you distain of the term GNU/Linux, I take offence at being called brainwashed. I’m not a Linux “zealot” nor am I intrested in telling other people if there right or wrong. I am simply stating my opinions on the matters being discussed.
Intel is coming with processors for high-end machines. You can easily port Linux for that.
It’s called IA64, and the port has already been done…
Once Linux becomes better other solutions will become irrelavent.
Linux has a long way to go… the 64-way Altix system is AWFUL on I/O, both because of limits of its I/O busses and contentions inside the Linux kernel. Altix is being sold for the purposes of scientific research, which is its only domain as it doesn’t have the throughput for a high performance server.
The high performance server market will continue to belong to Unix.
Check out the latest profit figures from Sun. Sun is going down.
Well, yes and no. They reported a loss, but it wasn’t as large as expected, so the lastest news regarding Sun’s financial status was actually positive. It still doesn’t help the fact that their stock is at $0.14/share.
That’s why they are suing Microsoft, but they will not get anything, since the reason why Sun goes down is not Microsoft, but actually linux and open source.
Linux has no place in Sun’s market. Sun sells high reliability high performance servers, with one of the purposes (as I previously mentioned) being deployment of large scale thin client networks. There aren’t any programs on Linux which can even come close to touching the functionality of the Sunray Server (please don’t make yourself stupid and mention XFree86/LTSP) Linux runs primarily on low performance commodity x86 hardware.
Sun has even set themselves up as a Linux vendor. It doesn’t seem to be doing them a great deal of good…
I did not mean to imply that the license makes a good implementation. However, Operating Systems are now a commodity. There are free ones and commerical ones and they are all low cost. What seperates linux from the rest is that it works good and it’s licensed under the GPL.
And from what I’ve read before, RMS did a lot more than write an editor. A lot of people want to diminish his contributions because they have a personal vendetta against him. He _is_ the father of open source IMHO.
>Sarcasm on< What a revolutionary article. I had never heard about this up an coming OS. I was amazed to learn how this grass roots effort is going to revamp the entire industry by replacing the various flavors of unix. >/Sarcasm off<
Seriously, when are we going to see some real world studies that prove the advantages of swithing from Unix to Linux. I am very interested in running Linux in my company but we have a significant investment in HP-UX and Solaris systems. We have run several cost analysis to try an justify linux and I cannot find any real bennefit. Could this site maybe post some more “News” about real world adaptation of Linux vs Unix instead of re-hashing the same Linux is taking over articles. Just a thought.
Regarding the unity of the BSDs, I’ll give you that one hands down. So will everyone else (trying a link again) http://www.firstmonday.dk/issues/issue3_3/raymond/“> . There is no question that Linux is very unusual in attempting a project of this scope with as little organization or planning as it has.
Most of my systems have two completely disparate toolsets installed: tar and gtar, make and gmake, lex and flex, sh/ksh and bash, simply because GNU chooses to modify tools to the point that it breaks intercompatibility, rather than reinventing from scratch.
RMS was highly conservative about this as contrasted with the next generation. Linux generally has little problem breaking compatability as long as there is a version number change. Wait 10 years it will get much much worse. 🙂
You may want to take a look a fink patches as a long term reference on how to get Linuxy stuff to compile on a BSD box. They take their sources from Debian but OSX is BSD so…
At least with the latter I wouldn’t need to edit shell scripts which assume /bin/sh is bash,
Which I agree 100% is terrible. The two are not similar. I have no idea how this tradition developed on Linux.
go through C code and remove C++ style comments because they assume cc is gcc,
If you see this just change the makefile to use gcc. Save yourself the headache. Remember with OSX going to gcc as the system compiler this will also get worse as the years go by.
or replace all occurances of lex in a Makefile with flex because the author assumes lex is flex.
I gather this is a BSD lex?
I think you definitly have a point there. I have no doubt that an ISP can do no wrong with Linux but if you have a substancial investment in high end ‘NIX’s already, what would the benefit actually be? I have no idea what your company does but I take it your not using your ‘NIX boxes just for SMTP.
I guess I work with zombies then?
?
Well, yes and no. They reported a loss, but it wasn’t as large as expected
I remember this explanation, when market was going down last years, people were throwing out parties because their company actually reported a narrower loss. Hehehe. There were even nice cartoons about this, where companies layoff lots of people, makes lots of loss and still happy, because they laid of people less than expected, and made a loss less than expected. ))
Linux has no place in Sun’s market.
Today, there was a news which talks about how Dell replaced Sun servers with their Linux computers. They used clusters. Hmmm, probably that wasn’t Sun’s market. )
Sun sells high reliability high performance servers, with one of the purposes (as I previously mentioned) being deployment of large scale thin client networks.
Google is a large scale think client network application. Guess what Google uses. Thousands of Linux PCs. ))
There aren’t any programs on Linux which can even come close to touching the functionality of the Sunray Server
Nobody here argues that Sun is worse in functionality, the problem is the value proposition, value proposition. The cost, the benefit. Linux doesn’t come close to SGI’s mainframes when it comes to functionality, but SGI is the one in trouble. What’s your point.
please don’t make yourself stupid and mention XFree86/LTSP
Luckily you already mentioned it. ))
Sun has even set themselves up as a Linux vendor. It doesn’t seem to be doing them a great deal of good…
It is too unfortunate you weren’t able to understand and analyze why Sun is offering a product which really doesn’t make any sense at all for them. You yourself makes the point that it is not doing any good for them, so why the hell Sun offers Linux. Publicity, Publicity. Sun is now totally playing with emotions and these are their effor to be seen as a good company, since it seems that Sun is betting itself on the court against Microsoft. Eugenia had a great article about Sun’s court. Check out that.
RMS was highly conservative about this as contrasted with the next generation. Linux generally has little problem breaking compatability as long as there is a version number change. Wait 10 years it will get much much worse. 🙂
This is well and good on a Linux system. Most of my comments here have to do with compiling software developed with GNU development tools under Solaris.
You may want to take a look a fink patches as a long term reference on how to get Linuxy stuff to compile on a BSD box. They take their sources from Debian but OSX is BSD so…
I’m not sure if you saw my comments about OS X before or what, however in this case I am talking about Solaris.
“go through C code and remove C++ style comments because they assume cc is gcc,”
If you see this just change the makefile to use gcc. Save yourself the headache.
I really dislike using gcc on sparc64 unless absolutely necessary. gcc is literally at least 10 years behind the Sun compilers in terms of the binaries produced. My policy tends to be if more than two or three tweaks are required to get a particular package to compile under Solaris it probably isn’t worth using.
That said, I find qmail a wonderful choice for an SMTP server, and if you don’t want to purchase a license for the Sun ONE Message Server (a.k.a. EdgeMail) the UW SSL IMAP/POP3 daemons make a suitable replacement.
I had previously attempted to build the Courier IMAP daemon because I thought it would be nice to have an IMAP daemon with native Maildir support, however Courier is tied at the hip with GNU development tools (its docs explicitly say not to use the Solaris make)
So, I’ll stick with what builds more or less out of the box, as I don’t want to rely on any software I’ve had to hack a great deal to even get to compile…
Ehhhh… interesting pattern: as long as there is no proof to be found, -x*Linux*x- is the Best Operating System and will kill and stomp over all else….. but as soon it’s shown that it’s inferior to, for example FreeBSD, then, uh, “that benchmark isn’t important anyway, ‘coz in 2.5.66787 things are completely different, and -x*Linux*x- will the Best Operating System and will kill and stomp over all else Real Soon Now.
But regardless, I am sure there is no Linux zealotry involved here.
Lol! A very good point. I was working with creative labs when XF86 4.0.0 came out and had do speak to many open-source “advocates” for driver reasons. Mostly they were Obnoxious people with flimsy arguments. Although I stand by my comment on the userbase being Linux’ major asset, the nuts really don’t do the os any good at all.Most people lothe getting into discussions with them couse they generally get flamed.
“Sun sells high reliability high performance servers, with one of the purposes (as I previously mentioned) being deployment of large scale thin client networks.”
Google is a large scale think client network application. Guess what Google uses. Thousands of Linux PCs. ))
Umm, perhaps you don’t understand what I mean by thin client:
http://www.sun.com/products/sunray/images/sunray1.jpg
This is a Sunray. It’s somewhat similar to an X terminal, except it provides quite a bit more in terms of session management. First, there’s no need to login with a username and password, just stick your Smart Card into the reader and it can pull up a new or saved session for you. Need to move to a different location? Pop your smart card out and stick it in another Sunray, and your entire session will be seamlessly transfered.
Because Sunrays are so simple, they’re extremely low maintenance. All the maintenance that goes on happens on the server end. Contrast this to the typical workstation setup, in which each user has their own complete system with its own particular software configuration. A team of SAs is required to run around to all of these machines whenever something breaks, and problems must be troubleshooted individually. That’s not to mention the power consumption difference between a complete PC workstation and a Sunray, especially the Sunray 150 with an integrated LCD:
http://www.sun.com/products/sunray/sunray150/
Sun has the right idea with Sunray networks, the problem is a lack of software (apparently StarOffice doesn’t cut it) and the high cost of deployment, which seems to be too much to offset the reduced TCO.
Hello,
People read this link:
http://www.serverwatch.com/news/article.php/1399111
When Solaris Intel source code will be available than soon, after months or year it can be dominant over Linux!
Now that’s a kick in the pants to Randy.
Linux is new, its trendy
*chuckle* If I got a dime for every time I’ve seen this one the last seven years (which was when I traded my Amiga for a PC running Linux)…
As for the rest of this pissing contest, I think I’ll give it a miss.
“It’s certainly true that Linux does not obliterate Unix. Solaris has Linux beat hands down in terms of raw throughput,”
[i]Actually no, sorry ain’t true anymore. The http://www.sgi.com/servers/altix/ fastest computer in the world is a Linux box.
Fastest computer in the world? According to http://www.top500.org/ the fastest supercomputer in the world is the NEC Earth Simulator.
64 Itanium2 processors per node, each processor supporting up to 8 gigs of very high speed ram; disk control subsystems allowing reads at 2gBytes / second. Sun doesn’t have anything in that speed range.
You’re confusing memory bandwidth with I/O bandwidth. The Altix has shitty I/O, that’s why it’s designed specifically for CPU bound tasks such as modelling, rather than I/O intensive tasks a server would normally deal with.
A Sun Fire 15k has a peak system bandwidth of 172.8 GB/s, and a sustained bandwidth of 43.2 GB/s. Its I/O bandwidth is 21.6 GB/s sustained.
Sun has been the king of I/O for several years and I don’t see that changing any time soon.
Read the numbers and weep…
Sun has never been the king of I/O. SGI has always had them beat. Sun has always good I/O, but that does not a king make. What Sun has always offered was a balanced package at an attractive price.
No, SGI’s market for their supercomputers has always been scientific. SGI does not sell servers, they sell high performance systems for handling highly CPU bound tasks, not I/O bound tasks.
Once again, Jesus Christ, get a clue…
Umm, perhaps you don’t understand what I mean by thin client
I didn’t know that you specifically refer to smart cards and the related technology, but it is the same concept and you didn’t say something that will make me think about this particular technology. Thin client means something like a dump terminal of the past.
Forget about all of that, but here is the deal. The increase of the speed in network technology and the mainframes made some of us believe that we are going back, where we will have dump terminals and all of the applications will be hosted on the central computer.
Although there are advantages to that, still the smart client wins. Here is why? A fast network with a smart client is more powerful than a fast network with a dump client. We need processing power on the client, not on the server. X Windows was a research idea, and a good idea, but it is not the right solution for us. People do not want to use dump terminals.
Sun has the right idea with Sunray networks, the problem is a lack of software (apparently StarOffice doesn’t cut it) and the high cost of deployment, which seems to be too much to offset the reduced TCO
Sun doesn’t have the right idea, and I think they never did, except Java and the time where workstations did matter. Now, we have powerful PCs.
I don’t want my PC turn into a dumb terminal just to make Sun richer. This is not a dumb and dumber movie. We want our PCs smarter.
By the way smart card idea reminds me the floppy age. Why the hell should I carry my session with my smart card, when there is network. Store my session on the network, and retrieve it when it is necessary. Why do I need smart card to store my session, and at least I can remember every detail and restore my desktop if I need to. )))
thin clients…what about Citrix XP Metaframe, its what I use.
I have a “dumb” terminal Pace Box, hooked up to a LAN, its OS is RISC OS, but if I want to use any MS application, it just fires up Citrix Metaframe XP, and gives me a session to Window 2000.
However when I want to use the internet, it goes through a SUN server.
Also I when I registered for this system, it was through a Linux server. Interesting hey.
Sounds like Sun, are offering a total package although, what have they done to market it? they should market it, without a doubt this is the way forward especially where I’ve worked.
So Linux is popular because it’s targetting the desktop. Big deal. This may be exagerrated, but if Linux keeps on fighting for the desktop and doing features, it’ll soon become a second ms windows in terms of reliability, security,…
Features and reliability/security just do not mix.
Some people will prefer these funky features, but I (and I think I’m not alone) don’t. That’s why I run OpenBSD, which runs all the apps I need (I don’t need to be able to play *insert*latest*3d*shooter*; I couldn’t care less), and does so on a solid, fast (linux is _slow_ compared to my OpenBSD system; mostly because it runs all kind of shit by default), and most of all, secure foundation.
So, in my opinion, Linux is becoming a mainstream desktop OS, and is losing the high-reliability market. So when will Linux-viruses become common?
one may correct me if I am wrong….
A thin client/server solution couples relativley inexpensive(albeit not cheaper than Walmart PC’S with linbloze)”terminals”(for lack of better word, for modern thin clients are much, much more than terminals ever were) with the large scale computing power of a big-ass server. LTSP turns inexpensive PC’S into thin-clients. Smart clients allow for some local computation and execution on the clients but allows for system wide control from the central server.
Workstations and PC’s are two differnt beasts- Workstations are built to be application specific, allowing for a greater degree of optimization in terms of hardware and software for the application at hand- pc’s are general purpose microcomputers which do a little of everything and nothing particullarly good. Dedicated hadrware/software solutions are still and will probably always be the best bet for specific dedicated applications.
Linux presents something new to the pc world in that it allows for degree of specific application implementation which other OS’s have not provided before. Linux based pc’s can be used as work stations, as thin clients and as smart clients, and by virtue of the fact that it works on most hardware it is a cheaper solution than anything else due to the economics of mass-scale manufactured hardware and its (linux) non-existant cost.
In a corporate setting most of the goodies available on modern pc’s is fluff, likeable, perhaps enjoyable fluff, but fluff nevertheless. In the long run it will still be cheaper to produce a limited number of really powerfull computers(which cannot be mass produced) coupled with truelly high speed networks (ie. +1Gigabit/s) connected to “smarter” application equipment than the current approach of trying to saturate the computing market with budget PC’s.
The PC, as we know it, will not vanish anytime soon, but it will probably become irrelevant except for enthusiasts and developers/producers. At odds here are two factors: what corporations want and need in terms of computing power for their workforce and the enabling power of PC’S as devices which enable indidividuals to become independent producers. Neither approach will win, both will persist-for there is no direct competition between the two worlds.
With the advent of the internet, in its present form, the individualistic dream which came to fruition with the PC has given way, at least partially, to the development of vast decentralized networks of people working together independently -this development contradicts at once the corporate computer usage paradigm and the individualist computer paradigm, allowing for a degree of individualism and collectivity, and this is where Linux really shines and where all other OS’s pale in comparison.
Unix will not completely vanish any time soon. As long as large corporation need to provide their workforce with access to stored databases of information, Unix will survive. Linux will replace many current uses of Unix, but not all. linux is NOT Unix, it is the best of the Unix world, ie. its standards, its networking capabilities, etc. tied to defacto industry standard mass-produced hardware. Linux can run on anything, from a cell-phone to a mainframe, but its real advantage is tying popular home pc’s into the networked world of computing resources.
The desktop is but one of many applications for which linux can and is being used. Linux is already suitable as a desktop alternative for a certain segment of the computing population, like the word “alternative” itself suggests. Unix was not and is not intended for desktop usage. Other OS’s like winbloze and BEOS only cater to the media consumption/producer paradigm of computing, they are not powerfull enough to meet the large scale networking demands of corporations. They are soley based on the individualist paradigm of computer usage. If OpenBEOS goes the path opensource/free as in beer it may become a more user-friendly alternative to linux in the development of networked communities, but by the time it gets here, linux will be much more user friendly than it is now.
And of all the OS’s which exist linux is the only one which trully addresses the desires and demands of the people who started the microcomputer revolution, and a revolution indeed it was-yet as with most revolutions, they themeselves are subject to revolutions-the consumer paradigm was and is a counter-revolution, a revolution against the spirit which gave birth to the microcomputer. The consumer/producer paradigm is inherently egoistic and utterly dependent upon massive corporations which produce the hardware and software which enables it to be, which is why intellectual property and digitial rights is the alpha and omega of the winbloze world.
My how things have changed since Wozniak stole some TTL components from the stockroom at his employer (Hwelett Packard)and coupled that with a Mostek(which was formed by by ex-motorala 6800 engineers who Jobs had convinced to quit their jobs and form a new company) processor to build the first Apple…..
So Linux is popular because it’s targetting the desktop. Big deal. This may be exagerrated, but if Linux keeps on fighting for the desktop and doing features, it’ll soon become a second ms windows in terms of reliability, security,…
I totally agree with this statement.
Linux will may be the most popular desktop OS, but just after becoming a second windows. It’s on the right way to this “transformation”. And because of this, Linux will never kill “real” UNIXs.
I think you need to try 2.5.x Linux kernels, they’ll eat your heart out. Compared to 2.4, 2.5 blows the doors out of it. Linux 2.6 will be what FreeBSD 5.0 is with more.
Try 2.5.x kernel if it boots for you, those numbers so wrong now.
Try 2.5.x kernel if it boots for you, those numbers so wrong now.
I’ll wait until it’s released, thanks.
Will the lot of yee grow up, your like a bunch of toddlers complaining about who has better toys
toddler1: Linux rocks
toddler2: no use FreeBSD
toddler3: BeOs for ever!
Both Linux and the *BSD’s are opensource, they all have their strengths and weaknesses. I use Redhat but FreeBSD sounds intersting, but all the constant trolling and zealotry is just such a turn-off. If the OS does the job for yeah fine.
Bascule you have any plans on publishing your test program? sounds rather interesting could be of use to both the developers of Linux and *BSD (i hear Contest has being ported to FreeBSD). Might be interesting to also post in your benchmarks results for FreeBSD4.* as well. That way we could see improvements gained in FreeBSD5. As for Linux2.5 well development has slowed at the moment as Linus has gone on holiday so you could always test 2.5.59 or Andrew Mortons patched kernel (alot of the VM work is tested in it before being sent to Linus)
“GPL is the most empowering license ever devised and that is why Linux will succeed where others have failed. Thank God that RMS had the foresight to write software released under the GPL.”
And the GPL is also the reason why my employer will not allow us to extend any GPL based code.
We are allowed to use GPL code that are tools, but if it is code, we can’t touch it unless it has a license that allows us to extend it (i.e the Apache license)
>> linux is _slow_
Like said before, Linux 2.5 is a huge improvement!
Everytime I’m booting into Linux 2.5 I have to load different settings for my mouse speed because it is much faster when using 2.5
(just a small example, lots of other things are faster too)
And 2.5 makes everything feel more polished too..
This article should have never been posted. This is sort of like a troll, and Eugenia is enjoying her audience with this little war when in she knows for a fact which is far more superior. Unix…of course, silly.
so someone made a free unix clone..
who’s gonna make the free windows clone?
I think that Mott is right. Linux and BSD has hurt the commercial Unix market far more than anything. The effect was pretty fast in x86 land, where commercial Unix is now all but dead. I think this happened because the commerical Unix OSes, in general, don’t offer anything unique enough to warrant a purchase over the free Unix OSes.
(This also is the core reason why SCO is investigating the IP issue, I believe. The free Unix has gutted their Unix customer base.)
However, even if all commercial Unix vendors go under today, I will not cry for them; no more than I do for the dead DOS or Atari markets. Things change and nothing lasts forever; not even Windows.
A lot of people like to get into their little camps and talk about which UNIX is better. I really don’t care about that as I have used them all and they all have good and bad aspects. The good thing that Linux brings is adaptability. Despite the “poopyness” of the operating system it has managed to adapt to fill address real needs. When you compare it to a very mature system like Solaris, sure Linux will fall short in a lot of areas (preemptive kernel, process accounting, etc). But with companies like IBM and Red Hat offering comprehensive support for Linux and dumping huge amounts of money into development the gap between Linux and the “real” UNIX systems is really beginning to close. Although, it’s pretty safe to say that commercial UNIX as we know it will not simply disappear off the face of the world it’s a pretty fair assessment that Linux will continue to erode the market share of companies like Sun.
Its not always a question of what is better from a technical standpoint. You can talk all day long to with you computer science professor, who by the way has probably never worked outside of the university, about how the inner workings of BSD are far superior to that of Linux and because of that everyone should us BSD but at the end of the day who cares. Based on the opinions of most people on this forum I would say Windows is much more “poopy” than pretty much anything out there but it enjoys a huge market share.
Both Linux and the *BSD’s are opensource
I think that’s the rub right there. As long as there are people who like hacking on them they’ll never be dead. It’s only dead when it’s abandoned. These projects exist not because the market is screaming for them or because they’re trendy, they exist because there are people out there who like working on them in their spare time.
Sure, they’re getting popular now because they’re trendy, but just because one or the other isn’t popular with the business community is no indication of their “life”. They could be totally snubbed by “the market” and they would still exist simply because people like working on them.
Look at Atheos. Is there a market for that? No. Then is it dead? No.
Heck, DOS should be dead, right? I bet these guys beg to differ: http://www.freedos.org/
Heck, DOS should be dead, right? I bet these guys beg to differ: