What is the current status of operating systems as far as web hosting goes? Here is a quick run down:WINDOWS
As Microsoft has aggresively entered the server OS market through the popularity of their desktop operating system market they will have a tough road ahead. Microsoft has expanded its empire to video games, news media, the internet and more. Microsoft knows that windows won’t make them money forever. Recent announcements of major server providers using windows will help microsoft pressure users to buy the operating system. By offering their technology to major server providers at very low pricing and by offering many features that aren’t found on many unix operating systems they will not drop.
THE MONO/DotGNU FACTOR
.Net on Unix? Yes. This will provide an opportunity for windows users to use unix to its full potential while not having to switch. Since many Unix distributions are free users will now take a second look. With microsoft users being able to install a wide range of free software and the ability to run windows applications on unix will boost the usage of unix. Microsoft’s powerful market strategy is working and they won’t be gone anytime soon.
THE BSDs
The BSDs will likely increase market share greatly. With the announcements from SCO and tough pressure from companies wanting a more liberal license the BSDs (ekkoBSD, NetBSD, DragonFly BSD, FreeBSD, NetBSD, OpenBSD, MirBSD, OpenDarwin, Darwin etc.) will likely increase market share for 2004 especially after frustration of many server administrators after the latest linux exploits. Major server providers have also announced they will start supporting FreeBSD.
SOLARIS
Solaris, which is derived from BSD is famous for performing fast and being very secure. Solaris 10 for x86-64, x86 and SPARC will likely be a huge performance boost and will likely be adopted by many hosting companies. Solaris is the #2 Closed Sourced operating system used to host websites.
MAC OS X Server
What will happen to Mac OS X Server? Due to the inability for Apple to provide a version of its Mac OS System to other PowerPC OEMS it is unlikely that Apple will increase its market share in the hosting market. This would be the prime time for Apple to release an OS that is compatible with Mac OS X but not as “great”, therefore leaving apple as the crown jewel of computer systems. As all other operating systems tired to one specific computer system have failed or their market share has decreased greatly making the once-great company not worth as much as before.
LINUX
The recent attack from the SCO group will likely increase the market share of Linux due to the great publicity that has been generated, people are now saying “I heard of Linux, If someone is sueing $30 billion It must be great”. However, some Linux users may switch to BSD and Solaris in fear. Although some Linux users may switch their market share is still expected to rise. Windows Market share may rise in the short term from Linux users switching as we have noticed.
HP’s OS’s
HP’s server business has angered many guru users about discontinueing many valueable systems but HP-UX and OpenVMS now run on Itanium. Due to the disorganization that HP has caused users will be afraid that there is no long term commitment to continue providing HP-UX and OpenVMS. Many HP customers are expected to use Linux, BSD, or Windows operating systems.
The author basically predicted that everyone’s market share would go up except for HP’s and Apple’s. Does anyone else have a problem with that especially considering that HP and Apple are probably the least important players in the web server market (of those listed)? And I know this is only an opinion piece, but does the author have any qualifications for the predictions that he makes? I’ll stick to netcraft, thank you very much.
Solaris, which is derived from BSD is famous for performing fast and being very secure…
Solaris is based on SVr4 not BSD. You are thinking of SunOS up to version 5.
I’ve seen more informative, precise and thoughtful posts in OSNews comment sections.
Slow news day I suppose. No SkyOS beta .0.0.99 release to digest?
I disagree with a lot of the author’s points but I like the fact that OSNews lets people contribute articles. I learn a lot more reading comments by people that are correcting an author’s incorrect conclusions.
remember: bsdforums.org , this is a great site for help concerning FreeBSD.
It mentioned “NetBSD” twice in it’s list of *BSD operating systems. That’s about the only thing I liked about this article, in spite of it’s endorsement of *BSD. At least they picked a great operating system to repeat in the list. 🙂
There’s a lot of assumption in this piece. It’s a decent piece, but it should probably be an editorial.
I have to agree that I think Solaris use will go up after S10 comes out – looks like it can do some pretty wild things.
The bigger question is what constitutes use going up? I think that there is more and more web use, and therefore more and more servers needed. Windows 2003 will probably mean plenty of good things, but more than anything, it means people won’t migrate away as fast. I doubt W2K3 is attracting current *Nix/Apache peeps.
>Windows Market share may rise in the short term from Linux
>users switching as we have noticed.
Were/How did you notice that?
Do you have the source of this information?
This article is worthless. It’s all just conjecture with no supporting evidence and no real purpose. I guess OSnews needed some filler on a slow day.
Ok, if you make predictions, and publish it as an article, try to backup your predictions with decent arguments.. Currently it’s more of a combination of what you hope will happen, with a bunch of poofy woofy arguments slapped on it.
[i]Microsoft will let Mono/dotGNU stay compatible with .NET just enough until .NET usage becomes more than Java. Then, they will make .NET incompatible with Mono/dotGNU and ship it with Windows.[i]
Thereby making all commercial applications written for the windows .net also incompatible. Yup, that makes alot of sense. Especially from a company that has NEVER broken their windows api from the begining (not sure if that is Win95 or Win3.1) and will probably still be compatible with Loghorn.
Yup, makes sense for them to break it with .NET as well!
You know, one feature of Windows is that its filesystem is case insensitive. This means that you don’t have to worry about mistyping URLs. I sure as hell am not going to use it, though, but if I set up a webserver with UNIX, I would definately serve it from a Samba share, to get the case insensitivity.
Case insensitivity is good, I decided, when I tried to teach my 12yo sister how to program, and I made the mistake of picking Python.
“and the rest of the world will be fucked by another MS bait. again.”
it’s not bait unless someone takes it as such. Miguel knows he’s taking a huge risk with the port, and they will always be 2 steps behind MS…however this means *nix will be 2 steps ahead of where it’s at now. oh, and .Net != cashcow, it’s free (of course the OS to run it on is not, but win2k3 web server edition or whatever it’s called runs around 400 bucks, which isn’t too bad).
>> .Net on Unix? Yes. This will provide an opportunity for >>windows users to use unix to its full potential while not >>having to switch. Since many Unix distributions are free >>users will now take a second look. With microsoft users >>being able to install a wide range of free software and >>the ability to run windows applications on unix will boost
>>the usage of unix. Microsoft’s powerful market strategy is >>working and they won’t be gone anytime soon.
>Yup… Microsoft will let Mono/dotGNU stay compatible with >.NET just enough until .NET usage becomes more than Java. >Then, they will make .NET incompatible with Mono/dotGNU and >ship it with Windows. ISV’s will naturally then use >Microsoft’s version (== cashcow), and the rest of the world >will be fucked by another MS bait. again.
Doesn’t matter, mono is creating their own api set (GTK#) that brings the power of .NET to linux. MS would be crucified if they were to change their base API again after companies spent lots of money developing .NET 1.1 apps.
Umm, SCO are going after BSD after they are done with linux. ><
Case insensitivity is good, I decided, when I tried to teach my 12yo sister how to program, and I made the mistake of picking Python.
book book = new book(); // doesn’t seem that great to me
If you took the time to actually read the title, you would notice the OPTINION: tagline, damn troll.
When you don’t agree, you should write your own article with you own opinion. It is just too easy to attack somebody or some article by just saying “Lame article”. It says more about the commenter than the article.
so every time someone write a lame article I need to write a good article to justify my opinion.
I was told the article would be beefed up, but yes its opinion. It’s an article full of predictions from several hosting companies. Also, there is alot more Os’s than what is described above. AIX, IRIX, UnixWare, etc. Those OS’s are loosing market share greatly. The Internet is still growing 🙂 The great thing about this article is that users can trash the article all they want and users can correct mis information therefore making it useful in the end. I also do think It’s cool they actually posted this article on the site.
I got my source for the linux users switching to windows 2003 from netcraft.com. There are plenty of sources where I got the information. Next time I will spend a couple weeks on writing an article.. although this clearly isnt good enough to even be considered an article…..
“so every time someone write a lame article I need to write a good article to justify my opinion.”
No, sorry. You’d probably have to write alot of articles
. But comments like “Lame article” are not useful for an author, only insulting. If the person who wrote that comment went trough the process writing an article (and getting critique), he’d probably learn that writing decent comments is much more useful for both the author and the discussion, and less insulting to someone who invests his spare time in writing an article.
Many of the better features of BSD 4.3 UNIX were built into SVR4. Ultrix, SunOS 3.X and 4.X, Solaris 2.x, and SVR4
Solaris – (Sun Microsystems) Sun Unix-based user environment, including the Unix operating system and an X11-based window system. Solaris 1.x is a retroactive (marketing?) name for SunOS4.1.x, a BSD-like version of Unix with some SVR4 features. Solaris 2.x (which is what most people mean by “Solaris”) includes SunOS5.x, which is an SVR4-derived Unix.
Come to think of it-Isn’t there alot of BSD stuff build into the linux kernel? The GNU System? Isn’t there BSD in all unix?
It really does spark the debate on what BSD-derived means, now dosen’t it.
I was told Solaris is a combination of BSD and Svr4
Okay, so if we are just putting out opinions here, which is fine, here is mine.
When thinking about web servers it is significant to think about what the servers will be serving. There are static text files and images – no server platform will win large market based solely on the ability to serve static files. This is not a selling point. That leaves us with other key technologies, most notably: ASP.NET, JSP and PHP.
ASP.NET is both really great and yet also very disappointing. It is great from the point of view of an old style ASP developer who got tired of processing submitted forms the old way. But ASP.NET is still not that great, it does not overcome the inherit problem – web browsers are a very limited way to create a user interface. You shouldn’t have to pull all your tricks out to make the interaction feel natural. You should have easy ability to show model dialogs. You should have powerful image processing controls built in. You should have highly specialized input validation done on the client side. Loading and saving data should be separate from loading the presentation structure.
ASP.NET and web browsers are a stop gap solution for MS. They will, when they have milked ASP.NET, reveal their master plan – which is to do Remote Avalon for UI w/ Indigo for data comm. Very little business logic would be in the app loaded on the client. The app would be served up from the web server and SOAP/Indigo would be handled by the web server.
The ASP.NET model of development is not extensible nor is planned to be the future. MS will either rope people into Remote Avalon or they will be stuck nursing ASP.NET to meet new features that it was never meant to.
JSP/JSF can be run on any platform, so the selling points will be on the cost to run the server. I’m not a Linux fan, but if I was a company and had to choose a web server platform I would choose Linux. The writing is on the wall, Linux is here and it is big. IBM, Novell, and many other players have made it clear.
PHP – not my thing but it is gaining steam, at least in my market area. I expect it is all over.
So I expect to see large growth in PHP and JSP/JSF. I expect to see med growth in ASP.NET. Since Remote Avalon is long off and will deployed on Intranets first, it doesn’t count in this discussion.
WINDOWS
slowing growth and then evening out
THE MONO/DotGNU FACTOR
picking up some of the ASP.NET points, but not a big contender
THE BSDs
PHP and JSP/JSF – slowly lose market share (BSD is my fav, but I’m just saying what I see)
SOLARIS
10 may be great, but it doesn’t really matter. The features of 10 are what you need on a ‘mainframe’. IBM will not sleep through another big loss to Sun, IBM can beat Sun on every point, even price if they have to. solaris will drop massively in the next few years, sorry Sun
MAC OS X Server
very minor growth
LINUX
major growth, especially in international markets on small servers. also on IBM ‘mainframes’
HP’s OS’s
dropping off
HP was never that big in the webserver arena. This has traditionally been suns play ground. Due to mismanagement by SUN linux and BSD solutions are kicking their butts. Still in discussing internet growth one needs to deecide what specificaly they are talking about. in the context of joe college student setting up a webserver on his DSL or a big business setting up a huge web based application environment.
Last I checked the big Unix web servers could do case insensitivity.
http://httpd.apache.org/docs-2.0/mod/mod_speling.html
” if I set up a webserver with UNIX, I would definately serve it from a Samba share, to get the case insensitivity. ”
Just serve the DocumentRoot from a mounted partition formatted with FAT-32.
This article is completely worthless.
quote: ” But ASP.NET is still not that great, it does not overcome the inherit problem – web browsers are a very limited way to create a user interface. You shouldn’t have to pull all your tricks out to make the interaction feel natural. You should have easy ability to show model dialogs… You should have highly specialized input validation done on the client side.”
you made a bunch of good points there. here’s a question to everyone – what would your interest be in a XUL based framework – yes, running only on mozilla – that lets you create front-ends to db’s, like in Access / VB / Delphi, with data-bound controls and grids.
So you get the benefits of a RAD, full, rich GUI like VB/Access, and the benefits of a web app b/c it is updated everytime you use it. i’d think something like this would be a killer app for Mozilla Firefox
Opinions are fine but even the opinions don’t say much. I don’t have to justify my opinion by writing another article either. That’s just rediculous. There isn’t even much in the way of relevant sources or reasons behind these opinions. A short paragraph on a few Operasting Systems isn’t even worth writing in my OPINION. In my OPINION it is worthless. It’s a two way street. If someone can spout off about their opinions with little too back them up I can surely voice my opinion. Besides if I have an opinion that says every operating system except Linux sucks without any supporting evidence I wouldn’t expect OSnews or anyone else to publish that and if they did I would be a fool to think that I wouldnt’ be shouted down on a forum like this.
i think your wrong about solaris… although sun has mis-managed their business but they are getting on track now. this sun burnout talk is stupid, they are going down yet they keep posting profits and keep signing more users. i would buy solaris to get all the cool things that come with it. to get the support. better than buying a redhat OS my opinion… now that i have posted my opinion i guess I should be sentenced to prison. this is the opinions from several hosting providers that have been in the business since the mid 90s.
I don’t think the Itanium2 port of OpenVMS is complete.
http://www.hp.com/products1/evolution/alpha_retaintrust/openvms/
I don’t know about HP-UX or Tru64 on Itanium.
Someone mentioned some other unices. SGI and IBM seem to be more into linux now, so Irix and AIX getting a little less attention. Unixware seems pretty dead, surprising since it was one of the first true unices on x86.
It’s actually the Itanium that is dead, Intel is switching to AMD64.
AIX is still healthy, IBM will release Power5 and Power6…
“Come to think of it-Isn’t there alot of BSD stuff build into the linux kernel? The GNU System?”
A Lot? Umm No. In a sense all *nixish OS’s have a common ancestory but there is not “alot” of it in the Linux kernel. Saying Linux is BSD “derived” is like saying FreeBSD is Eniac derived.
btw lame article. Sorry, no good suggestions for the author. This is an opinion forum not a place where I come to teach people how to do research. I or anyone else are free to say whatever the hell they want within reason and it doesn’t have to always be constructive.
Solaris 9 is svr4 based but the “Regents of the University of California” thing is displayed on startup so there is at least some stuff in there. I think you’ld be hard pressed to find a popular OS that didn’t have at least some BSD code though. Yes this does include Linux though I wouldn’t call it BSD derived per se.
Glad to see the author at least had the sense not to be one of those wankers that predicts linux will take over everything and didn’t diss solaris (sorry folks but Sun isn’t going anyware, though you don’t have to like it).
Later
Jared
“When thinking about web servers it is significant to think about what the servers will be serving. There are static text files and images – no server platform will win large market based solely on the ability to serve static files. This is not a selling point. That leaves us with other key technologies, most notably: ASP.NET, JSP and PHP. ”
You forgot one of the most important — Zope.
“Due to the inability for Apple to provide a version of its Mac OS System to other PowerPC OEMS it is unlikely that Apple will increase its market share in the hosting market. This would be the prime time for Apple to release an OS that is compatible with Mac OS X but not as “great”, therefore leaving apple as the crown jewel of computer systems.”
Er, isn’t that really the niche that Darwin fills? For the most part Darwin is just Mac OS X minus the Aqua interface and iApps, which really don’t matter much for a web server anyway.
As security is recognized as being more of a priority, I would expect that there would especially be gains in the Solaris, BSD, and Mac OS X area. HP-UX isn’t bad security wise but they’ve been working so hard at alienating their customers over the years I don’t expect them to reap any such benefits.
There’s also the price situation. MS-Windows (under current licensing schemes) tends to be more expensive than pretty much anything else when you actually price the whole thing with reasonable hardware, in spite of its ill-deserved reputation to the contrary. Solaris, the BSDs, and Linux can run on anything that MS-Windows can (and more in fact) and give a lot more flexibility with hardware; Mac OS X is somewhat restrictive with hardware (although Darwin can run on pretty much anything that MS-Windows can, too) but the price/performance of an XServe beats out equivalent MS-Windows boxes for any realistic use once licensing is factored in. In terms of maintenance, the BSDs and especially Linux require a bit more expertise than the others and Mac OS X possibly a bit less.
If you don’t think theres alot of BSD code in linux, then you must not be a fan of server administration
Thank you APW for your opinion as I agree with mostly all of it.
What the article does not cover is hardware, Dell/Linux is a much cheaper alternitave to Sun/Solaris. For this reason Linux is going to erode at Suns market share from the bottom up.
We all love BSD but it won’t be making much progress.
AMD and Intel are already moving into the server market, but they are going to make huge inroads when the 64 bit desktop becomes a reality.
Think 64 bit Xeon.
For this reason and improvements in Windows Server, I also believe MS is in a strong position if they don’t mess up.
Summary: Linux and Windows on Intel and AMD.
I very much doubt that Linux is going to stop increasing its market share. I don’t think anyone doubts that SCO is just dead. SCO has already said “we never said there was Unix code in Linux” after not being able to demonstrate it.
Linux 2.6 is going to rock in the web server arena. For one, the “anticipatory scheduler” introduced in the kernel 2.6 was designed with apache-like workloads. Just look at the original researchers: http://www.cs.rice.edu/~ssiyer/r/antsched/
“The Apache webserver delivers between 29% and 71% more throughput on a disk-intensive workload.”
“[…] especially after frustration of many server administrators after the latest linux exploits.”
1) Why should they be frustrated about local vulnerabilities? Sure, it’s not nice, but remote vulnerabilities or daemons which escalate to local user access are far more worse.
2) What did you meant? Userland too?
3) The amount of FreeBSD and NetBSD vulnerabilities in the kernel (in system calls) is about near the Linux kernel numbers. Last time i investigated over 1 year timespan, FreeBSD had even one more. It makes little sense naming EkkoBSD and MirBSD because they have almost no userbase. I feel left out as Debian GNU/KFreeBSD user!
4) OpenBSD had a few less vulnerabilities, but shared a common few vulnerabilities which were in all the major BSD’s. OpenBSD results in other problems depending on what it should be used for. One who wishes “secure by default” can tweak his/her userland and kernel at Linux to achieve a more secure OS. SELinux, PaX, GrSecurity, OpenWall, RSBAC, and MANY, MANY more. The possibilities are endless, see: http://www.adamantix.com as example of a hardened Linux distribution.
The argument is imo bull. Use what ye like, but this is plain illogic.
“what would your interest be in a XUL based framework – yes, running only on mozilla – that lets you create front-ends to db’s, like in Access / VB / Delphi, with data-bound controls and grids.”
I have thought about this myself. I recently did work in Enterprise Java where I exposed Session Beans as web services. Then I consumed those web services in XUL – using remotely hosted XUL. I had a little framework for ‘deserializing’ value objects (bulk beans) and all in all it was pretty cool.
I think that if only the Mozilla crew saw that the could be contenders against Remote Avalon they could create an strong competition for that space. As long as they think in terms of email/web browser then Mozilla won’t get the support it needs.
Here are my issues with XUL (although I really want to like it). I am suggesting remote XUL as the way to go because users and administrators don’t want to deal with installing apps.
1) remote XUL has many issues – by default you get all these popup messages about security when accessing your forms and your data. The support for localization depends on DTD, which is not available in remote xul – big issue.
2) XPCOM is modeled after COM – I mean query interfaces? really? I gave that shit up when I found Java, please a better object model
Especially since JS late binds anyway, why not bind against all interfaces (which we would in the Java class model or .NET class model).
3) you really are expected to know C++ when dealing with XUL, sure most things can be done in JavaScript. But if we want something that can compete w/ VB or Delphi it needs to have full and clean support in a language other than C++. It needs to be an equal language not a tag on, and I really doubt the Mozilla guys would go for that.
4) The support for vector graphics in XAML is just too sexy for me – unless XUL starts working this way I would keep thinking I was working with second class tools and my heart wouldn’t be in it.
5) The little research I did suggested there was not good support for image processing. For example, I want to load in a 20 MB image file and be able to zoom/pan that image very fast and have no flicker or artifacts.
Continuing both 4 and 5 I want a powerful 2D graphics library comparable to Java2D or .NET.
Dune_Buggy: I’m not really into data binding that much. I find it is always to simplistic and has a hard time w/ relationships. If you have a data binding solution that works for you that is great. In the bigger picture, though, XUL serves more as a proof of concept. Sadly MS seems to be the one benifiting from it the most. Hardly a week goes by that I don’t wish there was a system that combined the good parts of XUL and XAML and a few other odds and ends.
The most wise thing to do is to follow the leader, not ONLY because it is better in every regard but because it actually outsells the alternatives by a very large ratio. Java is here to stay, mainly because the most influential companies in enterprise computing want it to stay. Betting millions on a closed proprietary platform such as .NOT is fullishness, especially when you consider which company is behind it. I mean, come on ! Would you ever trust that company ? It sounds like a joke (how about a Micro$oft firewall ??).
Of course in our imperfect world political reasons play a strong role and .Not will have a percentage of the market. You could attribute that to M$ marketing but I just see inadequate higher education which produces incompetent IT consultants and biased customers.
“Umm, SCO are going after BSD after they are done with linux.”
I highly doubt there will even be a SCO after IBM gets DONE with them.
“It’s actually the Itanium that is dead, Intel is switching to AMD64.”
@ AMD fans:
Itanium is not going anywhere. Too much $$$$$$$$$$ dumped into it to just abadon ship now. Itanium still has a role to play for Intel. Check out the roadmap near the bottom of this this link http://www.anandtech.com/cpu/showdoc.html?i=1995 to see that Itanium still has life. The whole “Itanium is dead” comments are foolish just like the “BSD is dying” ones.
“What the article does not cover is hardware, Dell/Linux is a much cheaper alternitave to Sun/Solaris. For this reason Linux is going to erode at Suns market share from the bottom up.”
That’s completely false. Since Solaris can also run on the identical Dell X86 systems you seem to be touting, hardware is a wash. And with the various system administration tools available on Solaris (and not on Linux) it’s arguable that the TCO of a Sun Solaris system is less than the TCO of an equivalent Linux system.
“That’s completely false. Since Solaris can also run on the identical Dell X86 systems you seem to be touting, hardware is a wash.”
solaris runs piss poor without a sparc the last time I used it… so I believe his arguement to be valid.
hi
“I got my source for the linux users switching to windows 2003 from netcraft.com”
same source tells me that Linux users get a huge amount of windows switchers. So it goes both ways at the least
regards
Jess
Come to think of it-Isn’t there alot of BSD stuff build into the linux kernel? The GNU System? Isn’t there BSD in all unix?
One aspect of it I can verify – ipchains, which was originally ifpwadmin, which was originally taken from *BSD.
However, it has been extensively rewritten. This is because the Linux kernel isn’t the same as the *BSD kernel, and has a somewhat different focus. (Mind you, FreeBSD, NetBSD, OpenBSD, etc, are all *BSD, and they’ve all got their own particular focuses …)
Wesley Parish
Even though Dell is the most zealous (and one of the last still standing) wintel advocates and gets the best prices from its parental companies it is not the only option out there. Most people do not realise that they can build better and cheaper x86 peecees easily or even trust companies with better support and history at the enterprise.
If you want the ultimate combination and are prepared to pay the money you would be fool not to go Sun/Solaris. If you want just the best hardware, IBM PPC family is the one to get but you are stuck with Linux.
Solaris, which is derived from BSD is famous for performing fast and being very secure…
Solaris is based on SVr4 not BSD. You are thinking of SunOS up to version 5.
And you’re not thinking, SunOS 4.1.1 was Solaris 1.