“Is Server 2003 fast? Yes, it’s easily outdistances anything that Microsoft has ever shown us before. And, yes, it’s faster than Linux. It is stable? Based on about two weeks of testing with the final release, I would say that it’s also the most stable Microsoft operating system I’ve ever seen. But then again, I’m not asking it to do much besides basic file/print. The functionality servers that any business needs are largely not there. Ironically, this reminds me of the old claim against Linux that it didn’t have any applications. That was never true of Linux, but it is true of Server 2003.” Read the article at Practical-Tech by Steven J. Vaughan-Nichols.
Windows Server 2003: Incompatible, Fast & Confusing
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSAlert.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Well of course its fast, without anything installed on it it doesnt have the overhead that traditional Windows based OS of the past have had. Everything works great until you start to install applications on them.
As an operating system itself, it’s fine, but as the crown jewel of what Microsoft calls the “Windows Server System,” Server 2003 is zircon, not diamond.
Why? Just b/c Microsoft hasn’t released the other apps in the system doesn’t make 2003 any less of a jewel.
Secondly, why the is the author complaining about the lack of ICS and a firewall on the DataCenter edition? Your database should be behind a firewall, not running as the firewall.
If you can afford datacenter edition, then you have a complex security system set up. You wouldn’t even think about using Microsoft’s software firewal, or especially, ICS.
What? You mean a Windows OS is fast right after you install it? HOLY SHIT!!!
Server 2003 averaged 60% faster on file I/O than W2K and 80% faster than NT.
What registery condition, and what hardware, where those W2k and NT tests performed on?
ppppptttttt…. Wait till his registery fills up with all the non-sense that Windows pumps in there… Then we’ll talk performance agnaist Linux.
they should be faster.
that said, i’ve always been impressed with samba’s speed.
Most of the comparisons I’ve seen between Windows and Linux (in terms of Windows Server 2003 vs. Linux) are seemingly heavily biased towards windows… why you ask?
Take for example the fileserving benchmarks. They’ve taken Windows Server 2003, which I imagine has a native support for Windows Filesharing Protocols etc…etc…, and then compared it against Linux running Samba (an added program that make it possible for Linux to share files with Windows using their protocol). It just seems unfair that they are comparing Linux’s ability to handle Microsoft’s filesharing with Windows’ ability to handle Microsoft’s filesharing.
It says right in this article too that IIS 6 has a lot of it’s functionality built right into the kernel… so might we see similar improvements in the future with kernel mode applications under Linux. How does IIS 6 compare with Apache running in Kernel Mode?
I’ve not used Windows 2003 yet, however in the past I’ve found that Microsoft’s idea of filesharing is extremely slow and laughibly unreliable — it’s difficult enough to get Windows Explorer/Network Neighborhood to display the shared files/folders let alone access them. I’ve never had such problems after mounting directories over NFS. Which makes me wonder, is there any benchmark comparisons between SMB based filesharing and NFS based filesharing. I’ve always found NFS to be faster from user’s perspective… but who knows!
From an artical from the Windows Kernel team; the top requirements for Windows Server 2003 were:
1) Security (plug as many holes as posiable).
2) SMB sharing speed (it seams that Samba as faster).
3) IIS Speed.
Please note: the MSFT kernel guy even said that compatability was not a major concern. To get a lot of the improvements; the team just removed a lot of old code and actually removed a lot of obsolete APIs (normally the documentation said “don’t used obsolete” but MSFT has never removed any; DOS 7.1 still supported APIs that were obsoleted with DOS 2.0). Now, a lot of the older MSFT & third party applications will need to review their code and make sure none of the programs reference a removed API. Also note that they cleaned up security, thus some programs that needed extra security but were using a work arounds will most likely not work any more.
As a side note: Samba will be getting faster soon and should be on par with 2003 for speed. The lead programmer is in the middle of doing a complete code review and rewrite of the core system (it seams that things got a little messy with people just patching it to work with changes in the protocalls).
I have been running the Win2k3 server RC2 for quite a while now. Because of a few apps that wont run under Crossover Office I am stuck with windows one way or another. I use vmware to run one of my os’es on top of the other. I have done this both ways, with Redhat in VMWare on top of Win2k3 and the other way around. When running VMWare 4 on Win2k3 The system slows to a standstill no matter how low i set the RAM of the Virtual Machine. While on RedHat 9 Win2k3 runs just as fast in VMWare as it would on the raw hardware (or at least so close i dont notice the difference.) It really doesnt matter how fast Win2k3 runs on it’s own when it leaves so few resources unused that you have trouble running anything on top of it. FYI this is on a dual P3-1000Mhz system with 512MB ram.
What’s about the price?
Windows Server 2003, Enterprise Edition $3,999 US
…
Linux : FREE
Typical in that they continue to be THE worst company in IT for interoperability. They cant run their own software on their own damn OS…..let alone any other OS. As a result, typical that they keep everyone opening their pocketbooks to buy new software that will actually work on w2k3server.
Not typical of them making things more secure and faster. I welcome both these changes, but clean installs are always faster. Lets see how it does when the system has some stuff on it. Though the vs. Linux review was some what fair, its not the whole story. When it comes to serving up web, I see no test of the latest Linux kernel with pre-emptive scheduler and Apache 2.x. Furthermore, Id like to see w2k3server vs. a clean slackware box or even better, a uber optimized and stripped down Gentoo install. I’d put my money on the Gentoo box being really close…..and even faster once we get some apache optimizations into the Linux kernel. But lets face the facts here……If we bring $$ into the picture, a whole different story unfolds. Lets say you take a small fraction of what 2k3server costs and organization and put it towards some more hardware. Now set yourself up Linux/Apache cluster that would romp the snot outa w2k3server. And…..at the end of the day it still costed less.
Correction:SMB is a IBM protocol, not a Microsoft protocol.
IIS6 is build some degree into the kernel, why not compare it with TUX?
MS always tell people its latest version windows is xxx times faster than older vertion, but what is the platform? what is the system configuration? Of course new software can take full use of new hardware, like P4 hyper threading,etc.
Newer version not only make users buy once more, but also make ISVs recertify their software, of course that means money.
Something was a choice now becames a MUST.
MS always tell people its latest version windows is xxx times faster than older vertion.
Who knows if they do maximam optimizing for the newser one, but do some down-performance on the older one at the same time?
I find it quite interesting that this author has concluded that “Windows 2003 Server is faster than Linux”. He tested everything under a very small load (and without any performance tuning). All I can say, “faster in Linux” is a conclusion drawn out of non-realistic benchmarks which were COMPLETELY network-centric and limited to certain services (you cannot conclude an operating system is ‘faster’ than another by simply testing network performance. Do not forget some operating systems have certain hardware and disk space requirements).
Good try Steven, but I suggest you make some serious benchmarks against realistic server boxes (actually tuned for performance). I am not just talking web-server (same software, obviously….) and filesharing (this was interesting to hear, can be used against Linux, but this is one of the things. Try NFS….[which is better than SMB anyhow]).
Nice try but a redunant piece of work, mainly thanks to the ‘research’ backing it.
I must admit that I’m amused by all the posts talking about how it’s unfair to compare SMB speed. The bottom-line is that most business desktops run Windows, not Linux, and thus SMB is more important to them than NFS. Live with it. It’s not my Xbox’s fault that the Metal Gear Solid 2 port isn’t that great, but if someone I knew wanted to play only that game, it wouldn’t matter to him why the port is inferior; just that it is.
>>it’s faster than Linux
Based on what, wich application, wich hardware, wich Linux distribution? which setup…
… if you throw it from your window. “An OS has never been that fast!”
that will allow access to samba through explorer using a faster protocol? hell why can’t OSS come up with a NFS dll and then linx can just use NFS.
This doesn’t really sound good for Microsoft, and here’s why.
Microsoft’s main target for Windows Server Edition 2003 are NT4 shops. If anyone wants to debate that I can track down some links, but I think we are all pretty aware that this is the situation. Server 2003 is incompatible with the majority of the existing software (apparently).
The questions is this: If I am currently happy with NT 4.0, happy with my current applications, and have plenty of money already invested in my current infrastructure, why switch? Even if I did decide NT 4 was getting rusty, I would probably want to switch to something like Win2k so that I can keep my current software. I don’t think MS will get very far attempting to force upgrades on a company’s entire infrastructure, especially with the economy as it is.
Why pay for server operating systems? You can get N.O.S. for free(BSDs & Linucks) nowadays!
Come on. Why even post this?
let me guess? they’ve put print/file services into the kernel mode. i hear that the new IIS has many common HTTP responce paths in kernel…
i look forward to the howls of laughter as a certain spicy combintion of file, print or HTTP request …. as the kernel crashes 80% faster than before!!!
“Take for example the fileserving benchmarks. They’ve taken Windows Server 2003, which I imagine has a native support for Windows Filesharing Protocols etc…etc…, and then compared it against Linux running Samba (an added program that make it possible for Linux to share files with Windows using their protocol). It just seems unfair that they are comparing Linux’s ability to handle Microsoft’s filesharing with Windows’ ability to handle Microsoft’s filesharing. ”
It’s not unfair at all. It’s just a great example of why linux is not the best solution for companies. MS can provide 1 unified well working together system. Linux has to have a moreless of a hack to do the same. It’s not MS’s fault linux doesn’t have the same perfect support. Nor should it be their issue to make it just as perfect on linux.
People like having a out of the box, it just works approach to things. If one product was built to do the purpose and the other is built to try and do what the other is built to do, why would i want the later?
A comparision shouldn’t penilize windows for doing something natively because the competition cannot.
The only thing that Microsoft and Sun have over Linux is a middleware platform. All of the research and development that has gone into creating the 5000+ framework classes can be leveraged by corporations through inheritance because frameworks enable reusable design. Linux does not have these huge libraries. I would suggest that Open Source use Standard C++ Boost and other research and devlopment sources and build an open source research and development infrastructure. Linux could also have it’s own object-oriented platform (like .net and java) that businesses can leverage for solution implementation (as opposed to system implementation). This is my only concern regarding any situation where MS or Sun have an advantage over Linux.
With regard to Win2003, well it would be completely obsolete were it not for middleware which Linux has no clear answer for. The MS platform is obsolete because you can accomplish everything that can be done on MS on Open Source Linux. The speed differences are a non-issue because the differences are too small.
I’ve been running win2k3 enterprise for a few weeks now… and it’s great… really really fast (still doesn’t boot up as fast as Libranet though…) really really stable too… and is in fact much much improved over all the other incarnations of win in terms of memory usage… only eats 110m at bootup… winxp eats almost twice that at bootup… anyway i’m very pleased… all apps have installed and worked fine except for a couple, but this is remedied by running the install file under winxp compatibility…
anyway win2k3 is certainly the best windows ever… for whatever that is worth…
J Brown
now Wine will run more Windows programs than Windows can!
Absolutely not!
I don’t know what criteria the author bases his judgment on, but Windows 2003 Server is quite unresponsive with only a few applications installed on it. I haven’t done any official benchmarks on it or anything, but I have the exact same software installed on a 2000 server and a 2003 server and the 2000 server is more responsive.
J Brown, how in the hell did you ever manage to get Windows XP to eat up almost twice as much as W2k3 at bootup? That would mean a whopping 220MB of RAM going by your estimate, yet by all my experiences (Which includes a large variety of machines both at home, at school, and a few other places) WinXP takes more or less 100-110MB of RAM on bootup.
Win2k takes about 85-95 on my system, and all these figures are based on clean installs.
If you’ve got ANY operating system eating over 120MB of RAM on boot you’ve got a serious issue going on with your system, or major eye problems. Your whole post loses any credibility as soon as you spout anything so rediculous.
P.S. — Don’t ever expect any new Windows OS to take *less* RAM than the previous version. It’s foolish because it’s simply impossible due to the simple fact that SOFTWARE GETS BIGGER, especially such large pieces of software like an operating system.
After reading about the 2.6 kernel improvements in the May, 2003 edition of “The Linux Journal”, I think the next benchmarks after 2.6’s release will show something different entirely.
I don’t know what criteria the author bases his judgment on, but Windows 2003 Server is quite unresponsive with only a few applications installed on it.
Reconsider your criteria. When you talk about a server, “responsiveness” isn’t measured as the time it takes for a window to refresh or a widget to feel a click – it’s the throughput of its I/O subsystem. How many files can it keep open at the same time. How many sockets. How few page faults and cache misses while accessing cached data
A comparision shouldn’t penilize windows for doing something natively because the competition cannot.
Quit being ignorant. SMB is a proprietary Microsoft network protocol. Linux does have networking that works great out of the box with no add-ons. Of course some ignorant fuck-tard like yourself will think it is hack-ish for Linux to support above and beyond the native NFS support. I guess I can dock Microsoft for lack of out of the box appletalk… ass.
A comparision shouldn’t penilize windows for doing something natively because the competition cannot.
Quit being ignorant. SMB is a proprietary Microsoft network protocol.
See, this is what gets me about the Linux crowd. You can say “Well, Linux doesn’t do this as well … ” And then they come back wit “Yeah, but that is proprietary …..”
To that, I ask, who gives a shit if you’re in a business comprised mostly of Windows boxes, in which SMB is the protocal of choice?
If Win2k3 does SMB better than Linux, then what is or is not proprietary becomes highly irrevalent. Of course, if you’re outfit is comprised of mostly Linux boxes, then SMB is not going to matter. But if it does, and Win2k3 does it better, you can give excuses all day long about why Linux can’t do it as well (just like the font issue and the hardware support – it’s always somebody else’s fault, which is technically true, but not like anybody cares.)
So to the Linux pundets – Win2k3 does SMB better – every review I’ve seen so far says this. So, don’t go about bitching that the benchmarks are scewed or somehow completely off-base, or go on rants about the SMB protocol being proprietary and trying to give excuses for Linux. Just say that MS does this better and then go outside .. it’s probably sunny. I don’t care if SMB works better on Windows, and you shouldn’t either.
>> The MS platform is obsolete because you can accomplish >> everything that can be done on MS on Open Source Linux. >> The speed differences are a non-issue because the
>> differences are too small.
I’d be interested in seeing where you got the data for this comment. Exchange, for example, does not have a reliable, full featured, well integrated open source alternative.
I’ve worked with NetWare, Linux and Windows for a number of years and a lot of businesses choose Windows because it is well integrated, easy for their users to pick up since they have it at home/used in college/at their last job, etc and because of Exchange.
So to the Linux pundets – Win2k3 does SMB better – every review I’ve seen so far says this. So, don’t go about bitching that the benchmarks are scewed or somehow completely off-base, or go on rants about the SMB protocol being proprietary and trying to give excuses for Linux. Just say that MS does this better and then go outside .. it’s probably sunny.
*clap* *clap* well said.
tell ya what Darius…here is a fact……
Linux is 1 billion times faster than windows over NFS.
Linux is 1 billion times faster than windows over NFS.
Yeah, ok … so what exactly is your point? My statement was not an attempt to claim the superiority of Windows or the SMB protocol, so I dunno why the hell you dragged NFS into t his.
My intention was simply to laugh at the zealots who can’t even admit that Windows does at least one thing better. And even if they could/would admit it, they proably wouldn’t be able to sleep tonight because they’d be so pissed over it
Pathetic.
Actually…
Win2003 server may be faster than Samba. But I like to think of Samba as the SMB integrator. Various versions of Windows that have been left in the dust, or the old Domain system (which was replaced with AD) work just fine with Samba, and are depreciated or not supported with Win2003 server.
Samba is the clear choice if you have legacy machines. And what large company doesn’t?
“So to the Linux pundets – Win2k3 does SMB better”.
This is almost laughable. After all, who new that Samba would be a measuring stick for a performance comparison to Windows NATIVE cifs. What a joke!
I suppose that all of us (and I am a Linux user/advocat/enthusiast and ‘dows support tech) should be thanking Andrew Tridgell personally (along with all the other developers that worked on it) for shaking up the Windows world – thereby getting those inefficiencies addressed after it was long overdue!
ciao~
I don’t know much about Linux performance versus Windows performance myself, but I think I do not a bit about rational debate.
In the Linux fan crowd, there are some whingers out there who come across as a total bunch of losers whenever Windows v Linux performance benchmarks come off showing their beloved by-the-people OS as the weaker of the two.
Is it totally inconceivable that Microsoft could actually produce a better product? No. Not to any level headed rationally thinking person. There is nothing that says Linux is inherently faster than Windows.
So instead of bitching and moaning and slinging pointless ad hominems at any benchmarking organisation that dares to place Windows ahead of Linux, perhaps you should be pointing to benchmarks which prove the opposite?
Frankly, whether Microsoft is somewhere funding the report or not is not of great relevance. At most, you should be cautious in judging the report, but writing it off altogether just because Microsoft was lurking in the background is a completely irrational form of argumentation.
What matters is methodology – if you want to write a benchmark off simple because it had some funding from Microsoft, then sorry, but that is nothing but a lame, loser ad hominem retort.
In this case, the author of this article who knows far more about OSes and benchmarking than I do states his opinion that the Verifact benchmarks looked valid.
Mindcraft too copped a lot of argumentum ad hominem after their benchmark showed NT faster than Linux, so they did an open benchmark, it showed pretty much the same results, and consequently the Linux people suddenly went very quiet.
Step up to the plate instead of sitting in the crowd heckling – provide your own benchmarks, results, methodology. If the results are different then you can start discussing who has been cheating and how. All this childish “you got some money from Microsoft so therefore you must be a liar” is pathetic.
Grow up kids.
1st sentence should be “…but I think I do know a bit about…”
And Verifact should be Veritest.
Please Excuse any other poor grammar, spelling mistakes, etc.
Linux does come with an email server and also what about sendmail < http://www.sendmail.org >. I don’t think that there is anything that Microsoft offers that a Linux user doesn’t have. The Linux platform is also a secure platform. At any rate, what I value is open source projects because I feel that with Microsoft, people are forcing a lower quality platform to be the standard, and in order to accomplish that than repressive measures need to be taken that penalize the entire computer industry. I’d rather see dollars going into an open source model that is responsible to the public because nothing can be forced into distributions that can’t be removed if it is undesirable. In addition more types of development effort can be sponsored through open source, and not just projects that meet large business requirements. It doesn’t make much of a difference if one type of server is ten or twenty percent faster, or else slightly more efficient when you are forced to buy the whole package.
I mean if Windows couldn’t support its own file transfer protocols better than an open source competitor some people might start thinking there’s something wrong.
Windows should be more stable, more secure and faster than Linux. If it is not then it is yet another example of how short-sighted think-inside-the-box capitalists really are not the technology leaders they and their marketting departments profess them to be.
Microsoft compares to Linux like Bill Gates compares to Linus Torvolds. I don’t know what you think, but I think there is no comparison.
For once MS isn’t the bad guy here. Realistically, MS is finally doing what needed done a long time ago–cutting the fat. Apple does is every 3-5 years too and remains quite successful for it!
The whole marketing thing for MS is “easy”. In the past it was easy to convert from other stuff, easy to upgrade systems, easy to write programs, easy… Along with easy meant keeping lots of cruft around because users, corps, and espically ISVs won’t do the work to clean up their code for each version. That’s the real issue. In the change from NT to 2K most of the ISVs did the bare minimum to get the software to work rather than writing the software in the best available form. A good Win2K3 program should also be a good win2K, and winXP/98 program!
Three years ago Linux was at 2.2? Now if I had a linux binary program that was written then and complained that it didn’t work with 2.6 you’d all flame me for being stupid and not at least recompiling. [example: Where’s a good PHP4 plugin for Apache 2? OSS has the same problems too!] That’s exactly what the Cheepskate ISVs are doing here. They’ve had better than 6 months of RC2 beta to get patches out and haven’t. They probably want the $$$ for new versions (on top of maintenance contracts!) This is more a failure of MS marketing than Engineering. Of course if they were pushy about getting ISVs to comply then we’d complain about that!
Now the tiering and pricing is outrageous, monopolistic abuse! There’s been a stedy migration of features from the main OS to the servers. Lately, you can’t even run external non-MS server software (I use WinProxy as a firewall) on a win 2K server box without paying the outrageous CALS even when you’re not using MS services. But that’s another story.
I’d wouldn’t be looking towards Windows OR Linux if I wanted a secure OS. Linux is not the answer to that problem.
Re: James
This is almost laughable. After all, who new that Samba would be a measuring stick for a performance comparison to Windows NATIVE cifs. What a joke!
Yes, that’s my point .. so why are Linux pundets trying to argue about it? It’s like trying to justify why MS Office runs better in Windows than in Crossover Office. Hello? The reasons should be obvious to anyone, or at least those with no political agendas.
Re: Anonymous (IP: —.cg.shawcable.net)
Linux does come with an email server
Linux doesn’t come with shit except for the kernel.
and also what about sendmail
So, what about it?
I don’t think that there is anything that Microsoft offers that a Linux user doesn’t have
Perhaps true on the server (as this discussion has centered around), but I dunno … definitely not on the desktop though.
The Linux platform is also a secure platform.
Keep in mind that now you’re comparing Linux on the server to Windows Server 2003. Not saying that Win2k3 is secure or not secure … only time will tell.
The Linux zealots who have never sat on any standards meetings need to clue in to the fact that he who has the most market share for his technology/protocol/whatever becomes the de facto standard, no matter if it is proprietary or not. Point blank, end of story.
See, this is what gets me about the Linux crowd. You can say “Well, Linux doesn’t do this as well … ” And then they come back wit “Yeah, but that is proprietary …..”
To that, I ask, who gives a shit if you’re in a business comprised mostly of Windows boxes, in which SMB is the protocal of choice?
Not all businesses are mostly Windows boxes and SMB isn’t always the protocal of choice.
If Win2k3 does SMB better than Linux, then what is or is not proprietary becomes highly irrevalent.
If Win2k3 does SMB better the article should say it does SMB better as opposed to “It is faster than linux at filesharing”.
but not like anybody cares.)
I do. It really gets on my nerves when people use poor semantics and give the wrong impression. I don’t object to someone specifically saying Win2k3 has better SMB support. I do have a big problem with saying it is better at filesharing when the tests are done EXCLUSIVELY using SMB. It is misleading, ignorant, offensive to people who know better.
maybe the author has never used debian sid with 2.5 kernel
and some tuning, i don’t really think something can be that
faster, i just compare webdav as filesystem and linux beats
win2k3 any day of the week, maybe i can post a comparision using like 200 dual boot win/linux boxes and 1 server with win2k3 and debian. Webdav as filesystem is the future i think!!
Samba 2.2.5? That is about 5 versions old. Stable is currently 2.2.8a. And I have seen performance increases just from 2.2.7 through 2.2.8a.
Also, What kernel is Sco Linux 4 running? Redhat A.S.? Isn’t that an old kernel (2.4.5?). What glibc? Etc. (No time to check)
This is of course not comparing fairly. Win Server 2003 is the LATEST of everything from Redmond. Then this author compares to slightly older versions of Linux. This alone can make a difference. And where does he get these numbers? How do we know they weren’t pulled out of air? 55%, 110%, etc. Can we see actualy results along with how the benchmarks were setup/run? Again, can make difference. Also Some benchmarks aren’t even worthy of being a benchmark.
Anything else?
Actually any distribution of Linux comes with a mail server (Postfix/Sendmail), web server (Apache), file (Samba) and print (Cups), news, and boot (DHCP/NIS) servers, a firewall (Iptables), several database servers (Mysql/PostgreSql)…and the list goes on. You should try Linux some time.
I’m sorry, but MS has publicly stated that in the interest of the security of the OS, that they will not consider backward compatibility to be of much importance. Too bad, so sad. Shut up and buy the new apps when they become available, use the old ones on a Windows 2000 machine, or wait for a potential service pack.
Some of the reasons these applications will not work properly on W2K3 is because of changes to the AD Schema, with Exchange 2000 being the primary example. HOWEVER – Exchange 2000 running on a Windows 2000 Domain Controller in a Windows 2003 Active Directory Domain will run just fine – even though it modifies the W2K3 AD schema. I suspect the same is true for other “server” products such as ISA, but exchange is the only one I have tested thoroughly enough to verify that it is the case without a doubt.
>> Actually any distribution of Linux comes with a mail
>> server (Postfix/Sendmail), web server (Apache), file
>> (Samba) and print (Cups), news, and boot (DHCP/NIS)
>> servers, a firewall (Iptables), several database
>> servers (Mysql/PostgreSql)…and the list goes on. You
>> should try Linux some time.
I have “tried Linux”. In fact, I run it in a production capacity at my job for web services because it’s the best tool for the job. That’s what this boils down to. What is the best tool for the job? It’s not always clear cut. In some instances, it’s an open source solution and in others, it’s a proprietary solution. Postfix/Sendmail don’t come close to providing the services that Exchange does which is my point. In some environments, they don’t *need* journaling, contact management, calendaring and other tasks to be maintained with a single client — mine does. In order to provide the best services to clients, you have to look at the broad scope of technology and pick the best based on functionality, cost, ease of use, ease of administration, etc. Cost is only one variable in the equation.
nuff said.
In addition, Linux is not just a server platform, it’s also a high quality desktop platform with the addition of desktops like Gnome 2.2 and Kde 3.1. I’ve been using Linux for the last 15 months at home and I’m now oblivious to any Microsoft technology simply because it is not necessary. Well it started for me as a developer because of the Windows system interface (Win32). It’s just an interface so it doesn’t allow me to work with the platform implementation. If I were using Microsoft technology, I would trade one interface (Win32) for another (.Net). There is no difference between one interface and another so long as you can interface with all of your hardware. With Linux you can work with widget toolkits like Gtk+/Gtkmm/Glib/Gdk and QT, but you can also work with the platform implementation, and this is where Linux is different by nature, it attracts system implementation while on the other hand, end user platforms such as Microsoft are exclusively dedicated to solution developers. Linux does support solution implementation as well through Java and Mono, but I would rather see an open source middleware effort with Standard C++ as the core langauge (Standard C++, not vendor C++).
My Linux desktop experience is good enough that I just don’t use Microsoft anymore, and if I can do it, more people are doing it. I was forced to use Microsoft when I entered the world of computers and it was awesome to be able to get that dirty animal off of my back. The Linux desktop experience comes with a sense of freedom.
The truth is that hardware vendors should be encouraged to publish information so that anyone can write drivers for the hardware. I would like to see an object oriented platform. Linux is a better choice than Microsoft for sure, but it would be great if developers were not held back from developing new platforms because an obsolete corporation who is taking a mighty fall is doing everything in their power to circumvent invention and degrade the industry.
The article probably has a point regarding the loss of backwards compatability (which is a good thing if it finally gets rid of the old APIs) and another point about the lack of current server software. However, his Linux comparison is meaningless.
Two points 1st about labs testing second about authors own testing.
1) Microsoft sponsored the research the author alludes to, so I wouldn’t trust the asertion that Win2003 is faster than RedHat Enterprise. I’m sure RedHat could sponsor a similar test in order to “prove” that RedHat is faster. The test was biased is my point.
2) The author carried out static and dynamic web serving tests. He talks alot about IIS. Dynamic webserving in a Microsoft context usually means ActiveServerPages. How did he test this on SCO Linux? He makes no mention of ChilliASP (or whatever it’s called) only Samba and Apache.
Scott Lowe, I won’t argue the idea that Linux can do everything better than alternative platforms, however, I think that if there is any functionality that is missing in an open source project than the implementors should be contacted and provided the new set of requirements. With the deployment of object-oriented design, it should not be a problem to integrate new functionality into a development iteration. I myself would not use proprietary software under any circumstances. I don’t understand why anyone else would. I wouldn’t be able to justify closed implementation because it’s apparent that all closed efforts are eventually discovered through open source if you just wait a while. Not only that, but the open source software is high quality, responsible, and reliable software worth waiting for.
I’ve been a linux user and Sysadmin for several years and if windows is faster than linux at file sharing so what. Kernel 2.6 wll be out in awhile and there will be more heated discussions. Let’s raise the bar and let windows have their day in the sun if the benchmarks prove true. I get tired of all the arguing just for the sake of argument. I loath the windows opeating system and microsoft just as much as the next guy but all the constant arguing and pissing back and forth is worse. Time will tell if it is even worth arguing about, in 6 months if there are 50 back doors in win2k3 then who will care that it can do samba twice as fast. And if IIS walks you right into the kernel and gives you admin privs then SMB speed really won’t matter.
Peace
My intention was simply to laugh at the zealots who can’t even admit that Windows does at least one thing better.
I take a lot of offense to being called a zealot and I wish you wouldn’t do that anymore. I know I am not always civil to everyone but I am always civil to you and such a label is not very polite in my opinion.
I like linux, but I am not biased towards it. I haven’t been using it long and I wouldn’t be using it if I didn’t think it was better. The one and ONLY problem I have with linux has to do with commercial companies ignoring linux in applications and drivers. In the future, I will buy no unsupported hardware and it will not be a problem. However, like I said, I have only recently started using it.
I am a zealot for liking it more? Am I a zealot for telling people I think it may interest what I use and why? You are welcome to say so, but quit referring to me as a “zealot” in public. I am not a zealot, just a satisfied user.
People are zealots because everytime an article is posted showing Windows as superior in some way, the Linux zealots jump on it and make excuses. When someone suggests the desktop is better on Windows, Linux zealots make excuses and then point to GNOME/KDE as “proof” Linux is there. When a benchmark is released, Linux zealots accuse it of bias.
They will agree with anything pro-Linux and disagree with anything pro-Microsoft. It’s pretty clear. They’ve been around their Linux buddies so long and been fed the propaganda that Microsoft = bloated, buggy, evil, that nothing will convince them otherwise.
Meanwhile in the real world (not OSAlert, Slashdot, or LinuxToday), Windows Server 2003 is making huge advances and Linux is still trying to play catch up in both the server, and even more so, the desktop markets.
And where does he get these numbers? How do we know they weren’t pulled out of air? 55%, 110%, etc. Can we see actualy results along with how the benchmarks were setup/run? Again, can make difference. Also Some benchmarks aren’t even worthy of being a benchmark.
You can go and take a look at the Veritest benchmarks for yourself, they were linked to from the article. Instead of asking whether they were made up figures or not, perhaps you ought to make the effort before assuming lies?
Microsoft sponsored the research the author alludes to, so I wouldn’t trust the asertion that Win2003 is faster than RedHat Enterprise. I’m sure RedHat could sponsor a similar test in order to “prove” that RedHat is faster. The test was biased is my point.
This is nothing but speculation and argumentum ad hominem. Until RedHat does actually sponser a similar test which shows that, you are just grasping at straws.
Why can’t you just accept that it could be the case that Windows Server 2003 is faster than Linux? Why do you choose to rule out this possibility? It’s natural that Microsoft having developed a product will want someone to do some indepedant tests to verify that what they believe about their product is true. It’s childish to assume that any company that takes a contract from Microsoft is so unprofessional that they would just make up the results. So instead of this silly nonsense, why don’t you either find a better source, a better methodology? Or at least provide some valid criticisms of the Veritest benchmark which prove it is totally and completely out of whack?
I’m guessing it is because there isn’t any such information that you know of, so you resort to slinging FUD straight back at Microsoft.
Again, grow up kids. It’s not the end of the world, and I’m sure in a wee while Linux performance will improve anyway. Don’t go losing any sleep. It’s just a server OS, not a religion.
“Typical in that they continue to be THE worst company in IT for interoperability.”
Yeah, right, like I don’t have to update all my system libraries every three freaking months to run the latest software on Linux. Library hell is a problem on Linux (DLL hell was abolished way back when XP came out). Talk about interoperability.
ignore Darius, he is just another windows whore. he’ll call anyone that sticks up for linux a zealot because like i said-hes a windows whore. he ignores the facts of the article and of the various points that he makes.. the very fact that Windows has to try to be faster than a hack of its own proprietary protocol is sad in itself. he obviously doesnt understand the effort that is involved in reverse engineering a proprietary protocol, let alone improving upon it.
microsoft is an illegal monopoly, its a fact, even proven in court; yet they will remain one and the only thing that is able to some-what tame them is OSS. if it werent for samba and other OSS projects then microsoft would have no reason to improve upon its protocol and various other parts of its operating system. im sure samba will find a way to outperform MS at its own game once again.. and im sure MS will come right back.
webserver in kernelspace == bad
Linux is 1 billion times faster than windows over NFS
That’s a pretty bold claim to make – where are your numbers ? (You *are* aware there’s NFS support – from Microsoft – for Windows, right ?)
Even so, you’d be nuts to use NFS for serving files to client workstations unless you absolutely had to – there’s a reason NFS is known as Not Fucking Secure.
Added to that, most benchmarks I’ve seen (regardless of who does them) puts SMB as the faster protocol. Unless, it’s NFS over UDP of course, but that’s hardly a fair comparison.
It’s this simple… people don’t know how to read. I’m going to use my previous comment to present this to you, because it seems to be what has sparked some of this, and it appears many are considering me to be a Linux Zealot.
Following is a breakdown of how these arguments take place:
— Author says “Windows Filesharing Faster than Linux’s”
— Author proves with “Linux’s Samba is now slower than Windows Native SMB support.”
— I say “If we want a true comparison for *filesharing* we’d have to take something more native to Linux, such as NFS, as using a biased method of saying who’s got faster *filesharing* by using a proprietary protocol is unfair”
— People say “You damn Linux Zealots, every time someone says Windows does something better you can’t believe it no matter what.”
The argument here isn’t that Windows does or doesn’t do SMB based filesharing better, according to the benchmarks, it does. That’s cold hard proof that no one is going to deny. The argument is that SMB is not the be all end all of filesharing… if I were to create a network of my choice for MY company it would be all Linux, and SMB wouldn’t even come into the picture as it would be all NFS. To say Windows beats Linux in SMB base filesharing performance is one thing, to say Windows beats Linux in filesharing alone is something very different.
Is Server 2003 fast? Yes, it’s easily outdistances anything that Microsoft has ever shown us before. And, yes, it’s faster than Linux.
This is very broad and covers WAYYYY TOO MUCH. Remember the IPC Benchmarks from IBM that showed the limited Bandwidth of Win 2k and XP ? I do… here they are (http://www-106.ibm.com/developerworks/linux/library/l-rt4/?t). My questions to all you people claiming Linux Zealots can’t take a beating when it’s due is a) Has Microsoft improved on the speed and efficiency of much more fundamental issues b) Why don’t we ever see benchmark comparisons of NFS on Linux vs. NFS on Windows **oh, that’s right, I forgot… Microsoft is too busy innovating to support open standards**
In short… if we are going to compare filesharing as the broad term which it is, let’s do so. Let’s not say Windows is faster at serving up files with SMB therefore it’s faster at filesharing. And if we are going to compare IIS 6 with kernel level code (which you better believe increases it’s performance) let’s compare a webserver on Linux with kernel level code.
1) Large systems tend to use NFS because it’s an open protocall; MSFT doesn’t support NFS so you’ll need a 3rd party driver.
2) When running servers, it’s often nice to be able to run commands on a remote system. There’s an open protocall based around “rexec”. MSFT supports sending commands using this protocall but require you to install a 3rd party application to be a client. They also have created their own way that is not compatable with the open standard.
3) Why should I reboot the system just to add hard drives. My main computer at work doesn’t require me to do that.
4) MSFT’s engineers are using Linux as a point of comparison for what should be fixed in Windows. So far this has lead to several improvements:
a) Faster SMB sharing. It seems that Samba was faster and they didn’t like it.
b) They are working on improving the software update systems. It seems that they like Linux’s ability to do upgrades on live systems with out having to reboot. (Work under progress).
c) Most Web servers are running Apache, thus the reason for making IIS faster.
If it wasn’t for Linux, then Windows wouldn’t be improving as fast as it is. Also, if Windows didn’t exist, then Linux wouldn’t be improving as fast. The bigest thing to remember is that competition is good for everyone.
PS: Stop starting wars about which is better. If you want to help, you have two options: 1) Complain to MSFT and hope they fix it. 2) Write code for Linux and hope MSFT gets the picture.
Because if it isn’t I don’t care.
As long as Linux is relatively good at performance (and I think it will be interesting to see how the 2.6 tree competes, or how the 2.5.x does now) I will prefer it based on the UNIX design elements. Why can’t ms just admit they were wrong and create a unix based os like Apple did? Too much damn pride in redmond.
IF u change the login properties in WinXP by removing the welcome screen, u can boot your machine much faster and consumes less Ram space
People are zealots because everytime an article is posted showing Windows as superior in some way, the Linux zealots jump on it and make excuses. When someone suggests the desktop is better on Windows, Linux zealots make excuses
The “excuses” these “zealots” post are technical explanations for what you see in the benchmarks. It is an explanation of why things are the way they are. But thats all right, obviously your neanderthal mind has trouble grasping why they say more than “XXX suxxors”. It’s ok, it takes all types.
“the very fact that Windows has to try to be faster than a hack of its own proprietary protocol is sad in itself.”
I believe this covers my previous point which was directed to Darius… Nothing personal Darius. You’ve made some valid points…
ciao~
I say “If we want a true comparison for *filesharing* we’d have to take something more native to Linux, such as NFS, as using a biased method of saying who’s got faster *filesharing* by using a proprietary protocol is unfair”
The thing is, SMB’s faster than NFS as well – whether that SMB is being served from a Unix/Samba box or a Windows box.
Remember the IPC Benchmarks from IBM that showed the limited Bandwidth of Win 2k and XP ?
I do. I also remember a number of windows developers pointing out that what was demonstrated was largely worthless, because the methods used *wouldn’t* be the way you’d do it on Windows if you were “doing it properly”.
Why don’t we ever see benchmark comparisons of NFS on Linux vs. NFS on Windows **oh, that’s right, I forgot… Microsoft is too busy innovating to support open standards**
Not to mention that SMB is just better protocol for that sort of filesharing.
These days, NFS is largely restricted to sharing large chunks of data between trusted machines, not to serving out data hither, thither and yon to hundreds of untrusted workstations. As it should be.
On to the others…
1) Large systems tend to use NFS because it’s an open protocall; MSFT doesn’t support NFS so you’ll need a 3rd party driver.
No, large systems tend to use NFS because large systems tend to be Unix and that’s pretty much the only workable option you have – particularly to interoperate between between different platforms. This does not make NFS better, it just makes it more popular. There are vastly better options out there than NFS – it’s only used because it is the lowest common denominator.
Added to that, Microsoft do have an NFS module available for Windows, the Windows admins here tell me.
2) When running servers, it’s often nice to be able to run commands on a remote system. There’s an open protocall based around “rexec”. MSFT supports sending commands using this protocall but require you to install a 3rd party application to be a client. They also have created their own way that is not compatable with the open standard.
The r* tools suck. They are insecure and outdated. Don’t use them. Don’t suggest them. Just Say No.
Microsoft have “created their own way” because they looked at the alternatives, decided they sucked and wrote a better one. Just like Citrix.
I will never understand why some people consider the ability to admin their machines from a vt100 through an IP-over-carrier-pigeon link to be so important – and I’m a unix admin. Come ON people, drag yourself kicking and screaming into the 90s and embrace (and/or implement) some newer, better technology.
By Sylvain Giroux (IP: —.89-201-24.timi.mc.videotron.ca) – Posted on 2003-04-30 20:45:36
What’s about the price?
Windows Server 2003, Enterprise Edition $3,999 US
…
Linux : FREE
———————————————
Teknishn
But lets face the facts here……If we bring $$ into the picture, a whole different story unfolds. Lets say you take a small fraction of what 2k3server costs and organization and put it towards some more hardware. Now set yourself up Linux/Apache cluster that would romp the snot outa w2k3server. And…..at the end of the day it still costed less.
———————————————
I realize people around have this need to reflexively post comments around here, rather then letting thier cerebral cortex get involved. However, it gets really annoying to see X linux dolt, spreading stupidity around – especially since we all know, stupidity is more contagious than SARS. First off the price comparison. What business is going to download the isos of a standard distrobution and use that as their “mission critical server” ? Last I checked RH Advanced Server was NOT free. (ATTENTION: Sylvain Giroux, I repeat NOT FREE.) In fact its pretty darned expensive and close to the price of Windows 2003 Enterpise Edition:
Red Hat Linux Advanced Server:
$2499 Premium Edition
Windows 2003 Enterprise Edition:
$3999
So if we were then to follow some of the other bright ideas, and devote a little of this huge price difference ($1500) to buying better hardware. Since 2499 is a SMALL fraction of 3999. We could buy lots of lovely things, like decked out eMachine to run as cluster with our server? Because afterall the servers that Red Hat AS and Win2k3 EE run on are in that price range… doesn’t every companies IT dept. buy their stuff from Best Buy sales?
Then we have the lovely issue of support. The RH support period for the AS premium package is 1 years, yes a whole plurality of 1 yearS. Comes included with that 2499 sticker. Then there is the Win2k3 mainstream support, that is included with that 3999, you get 2 years. Now I don’t know where I can find out how much 1 extra year of AS support from RH is, I couldn’t find it on thier website – but I’m guessing that its not a cheap thing. I.e. it would turn that already huge $1500 difference for purchasing the uber eMachine into having the purchasing power for an uber rio-mp3 player.
I’ll bet all of this back-and-forth dialogue could be replaced by a few Linux textbots and a few Windows textbots, arguing it out. I think OSAlert should just replace the open forums with textbots. The result would be much the same.
Actually a lot of small businesses can meet information technology needs for $60 per box and free servers like Apache and mySQL, but you have to have a couple trained IT personel that know how to upgrade the distribution using source packages. I know of many computer enthusiasts who originally started out with X distribution version X, however it didn’t stay that way for long. If you are experienced under the hood, than Linux is much more cost effective.
I think that a large corporation should probably run their own in-house distribution. An open source platform allows you to have that much control, while with MS Windows you are bought into platform and have to pay for any sort of upgrade. There is no choice except to never upgrade.
People are zealots because everytime an article is posted showing Windows as less-than-perfect in some way, the Windows zealots jump on it and make excuses. When someone suggests the server is better on Linux/Unix/*BSD, Windows zealots make excuses
Define small business for me. As I recall, most businesses small or not, do not want to buy a pro-desktop version of software and then higher a full time IT staff to make it work like a server. Small business != small business + IT guy(s) and 1 RH Pro computer running mSQL/Apache/Samba.
If they want a server chances are they don’t want to “make” a desktop into a server. They want to buy somthing thats already a server, hard concept I know. Then there is this pricing thing, an IT staff is not your local linux dude going in and spending 5 minutes to fix a problem and getting paid 5.99 an hour. An IT staff is a dept. of people (usually more than one) that get paid on average about 50,000 a year if they are decently competent. Now which one sounds easier:
– buying a server, that does what you want and actually comes with technical support for server stuff
– buying a desktop; pretending its a server and buying a few really knowledgable IT guys to make it work
If want to say but RH Pro has tech support, it doesn’t have 24/7 phone support or any of the other juicy stuff that makes support work while. Furthermore its piddly support is only for 6 months and believe me if you were to call RH and say “yo I want to run mySQL for 50 employees on RH Pro and I got a problem” they are going to help you buy saying “we sell a SMB server version and it has technical support.”
Oh but wait you could say that you can use the LDP or the larger LDP we like to call Google. Then there is that small problem of trust. I have some dude named 733TLinuxH@kor posting packages on some strange website that supposedly fixes my problem. Do I trust them? How do I know it works? Even if you don’t take small business seriously, they take themselves seriously and no non-existant “real” tech support is not “good enough” for them. Nor is the the random tarball that claims to fix all their problems.
In conclsion stop it with the, “yeah I think 7 Eleven could manage all their systems world wide with 4 part time high school linux dudes, and maybe 3 downloaded ISOs of Mandrake 9.0.” Its not funny anymore.
All things about SMB being the wrong benchmark for meassuring network speed aside, there is still a problem. If you can produce the full specification of CIFS with no NDA attached to it and present it to the Samba team, I am quite confident that they would be able to make something as fast (or faster) than w2k3 server.
Reverse engineering a closed protocol is not easy work. Just be amazed that it even works at all.
I tend to agree, and I wasnt even going to comment on this until you did.
But to both crowds, network filesystems like smb and nfs are really .. old.
I’ve only seen smb used in schools and on student lan’s; at institutions, larger companies, subsidiaries etc. i’ve only seen fs’s such as coda and ncp (netware).
oh and this business with comparing the smb reverse engineered client for linux and the native closed driver for windows is in no way a good measure…
The Linux zealots who have never sat on any standards meetings need to clue in to the fact that he who has the most market share for his technology/protocol/whatever becomes the de facto standard, no matter if it is proprietary or not. Point blank, end of story.
Not true. Authorities set the Standards. People such as the government, ISO and the EU.
Otherwise it would be a case of “everybody drives down the motorway at 180mph so that’s the speed limit” or “Most gas installers are illegally working without a permit so that is the standard” No CORGI is the standard certification for a gas installer NOT what British Gas decides it is. Take phone calls, BT doesn’t dictate the STANDARD even though it has the biggest market share – OFTEL do.
Go work it out. Standards AREN’T set by what’s most popular, they are set by a set of open, published standards that anyone can work to.
Why can’t you just accept that it could be the case that Windows Server 2003 is faster than Linux? Why do you choose to rule out this possibility?
Because he doesn’t saw this with his own eyes, and who thinks that M$ is an angel good company, that’s a payed liar!
Why are you windows fanatics are so blind? Why cannot you accept the fact that Bill Gates and Steve Ballmer want’s your money and nothing else?
“Why are you windows fanatics are so blind? Why cannot you accept the fact that Bill Gates and Steve Ballmer want’s your money and nothing else?”
Why are you linux fanatics [are] so blind? Why cannot you accept the fact that Publix, WinnDixie, Albertson’s, and all the rest want[‘s] your money and nothing else? Welcome to business moron.
Ha, you MS zealots are so funny. He got you cornered and you ran out of excuses so you start flaming him. How pathetic and childish. It’s stupid elitist zealots like you that make me feel ashamed to be a Windows user!
The world doesn’t need elitist zealots like you! You do more harm to MS than good! Go away!
And not one (there might of been at the start but 81 comments later i’ve forgotten) actually bothered to comment on the article in a neutral manner, childish remarks at an all time high and zealots on both sides calling names.
OSAlert is about OS’ and where they are going, I don’t see why people see that as an opportunity to whinge and moan because they’re favourite OS of all those available which are catered for on this site doesn’t come top everytime.
Reconsider your criteria. When you talk about a server, “responsiveness” isn’t measured as the time it takes for a window to refresh or a widget to feel a click – it’s the throughput of its I/O subsystem. How many files can it keep open at the same time. How many sockets. How few page faults and cache misses while accessing cached data
Did I ever say that I was only talking about button clicking, scrolling and other widget interaction? No. I’m talking about the whole package. I’m talking about doing .NET development. I’m talking about serving large, database driven web applications. I’m talking about complex data transactions. I’m talking about the system as a whole. Windows 2000 “feels” better speedwise in all cases. As I said though, I have not done any official benchmarks on the two, so it is just my observation and opinion. You’re, of course, entitled to a different one.
Windows 2003 may be able to handle more files, larger loads, etc. However, most places windows 2003 will be used are not running into any of these limitations with Windows 2000, so to %80 ~ %90 percent of the people out there, 2003 will be a backward step into the swamp of slow. These are my experiences with the two systems as I implemented them. If you have different experiences with your implementations, then good for you.
“Because he doesn’t saw this with his own eyes, and who thinks that M$ is an angel good company, that’s a payed liar!”
I’m sorry, but am I supposed to believe that this is a rational refutation of the result provided by the Veritest benchmark which was founded by Microsoft?
Did you look up argumentum ad hominem in your dictionary or not?
“Why are you windows fanatics are so blind?”
Who said I was a windows fanatic? I pointed out the ridiculous ad hominems of a few nutty Linux nazis, not swear by my blood that Microsoft was the only source of good OSes in the world.
Pull yourself together – you’re not making sense.
“Why cannot you accept the fact that Bill Gates and Steve Ballmer want’s your money and nothing else?”
Errrr, that’s what any business wants – revenue. Providing a product that performs better than Linux say, is a good way of getting the money. Aren’t you just a genius.
and I’m not a tech, I was hoping to learn something. These open source guys seem to have a more flexible product, but the one point about meeting client needs determines what is best in each case is obvious. I am just desktop user so don’t belong here, but have always wondered whether to try using something stronger for networking. Running my own office forced me into a world of small business systems that are hard to envision. The article did sound scary in pointing out that bigger is stickier, I remember when microsoft fax actually worked on microsoft OS. I’m running W2K on three computers and when I look at my installed software, half of it is these constant patches.
I don’t know much about Linux performance versus Windows performance myself, but I think I do not a bit about rational debate.
I assume that last “not” was supposed to be “know”? But I really don’t think that’s true. Rational debate generally does not make use of the terms “crybabies”, “pathetic”, “childish”, and “kids”.
Question about RH Enterprise AS.
It’s not clear to me from RH’s site whether you are buying a one machine license or if you can buy one copy and install it on as many systems as you want. Can anyone answer this?
“People are zealots because everytime an article is posted showing Windows as superior in some way, the Linux zealots jump on it and make excuses.”
I’m sorry, but the reverse is also true. There are lots of Windows Zealots as well. Have you ever posted anything about Linux? Would you recognize that Linux is superior to Windows in some ways? If being an advocate for something marks you as a zealot, then you are one as well.
Bonch: “Library hell is a problem on Linux (DLL hell was abolished way back when XP came out).”
Funny, I’ve never experienced it. Dependency problems were abolished way back when modern Linux distros introduced intelligent package managers (such as urpmi on Mandrake).
I hate wasting my time reading articles by some pessimist who sets out to bash everything. This guy needs to be introduced to the REAL WORLD where other corps do the same thing, but don’t provide near the compatibility that MS does with most of their products.
Microsoft may have some issues, but I prefer security over compatibility. In addition, how about writing about some of Server 2k3’s GOOD POINTS. How about it? Unprofessional Journalism masked in good diction to prove a point – the author is a biased columnist who doesn’t like Microsoft. Load up anything else he writes into a wheel barrow and bring it to my garden for fertilizer.
If you patch your systems to rid yourself of one bug, you wind up with an entirely new bug. This seems to be a common theme with Microsoft — a company that can’t seem to understand why its customers don’t rush to install every patch.
Sorry bro, but that was just a big load. I manage Netware, Windows2000 Server, Unix (AIX), and Linux servers here. I guess Im confused as to why on earth you think you must go buy RedHat Advanced server for production environment. Thats complete non-sense. I need a Windows server…..I pay MS for it. I need Netware…..I pay Novell. I need AIX….I pay IBM. I need Linux…..YES, I do go and download the ISO and install it for free. WTF do I need RH AS and a big cost for? Support? I know linux and I support it….dont need RH. Just like I have never called MS for support on Windows2000. On a massive scale server such as a huge cluster serving thousands of people…..sure, RH AS or United Linux would be a better choice. Those situations are clearly not the majority though. You seem to forget that Linux is Linux. The major differences between RHAS and RH9 are going to be A) RH9 will have the latest bleeding edge stuff (potentially less stable packages and kernel) and B) RH9 wont include support. In general though Ive found even bleeding edge Linux to be far more stable than any Windows. It just sits there and does its job not needing reboots for everything. Its also curious to see that nearly all MAJOR installations requiring massive processing power in the world today use Unix or Linux….not Windows.
Regardless, Linux is still cheaper at the end of the day. And I wont have to worry about the next version of RH or whatever Linux I or anyone uses breaking compatibility with all my middleware. Win2k3 costs a lot and doesnt run MS own important software…either at all or stabily, but golly gee it sure is fast and it can beat an out of date version of samba…..neato!! As far as security in Win2k3…..we will have to wait and see, its much too early to tout that. As for Linux vs. Windows2k3 in performance. In a few short months the 2.6 kernel is unleashed and we shall see whois faster. (few short months…..thats fair considering it will be at least that long before MS even has a mail server that will run on its new uber server OS….perhaps SQL will be stable on it by then too) Looks like all its good for is being an overpriced webserver right now…..sign me up…bah.
And for the name callers….Im not a Linux Zealot. I use what works best for the job. As I said before, I admin everything from Windows to Netware to *nix. Funny thing is…..I use Windows for Email and SQL, Netware for file and print, Linux for all network services (web serving, all web based services, DHCP, DNS, etc), and AIX for our proprietary in house system. So lets see, based on what I use Windows for (as do many others)…..doesnt look like I even have an upgrade path now do I.
By Windows XP user (IP: —.upc-e.chello.nl) – Posted on 2003-05-01 15:22:24
Ha, you MS zealots are so funny. He got you cornered and you ran out of excuses so you start flaming him. How pathetic and childish. It’s stupid elitist zealots like you that make me feel ashamed to be a Windows user!
The world doesn’t need elitist zealots like you! You do more harm to MS than good! Go away!
——————————————–
First off, I prefer the terms Mac-Zealot, Apple-Fanboy, Jobs-Panzy, or if you must TiBook-Whore. Its a little difficult to call me a MS zealot when I don’t even own a computer that it would run on; but your welcome to try if you want.
Here’s a suggestion, if anything why not call me a linux community hater, penguinista molester, or if you must “really mean guy, who don’t like no linux peoples.” If you had bothered to read any of my posts you’ll notice I havn’t knocked linux itself, or the userland stuff sprinkled around it anyway at all. No SMB is cooler than NFS, no well linux blows because it hasn’t got better support for MS’ protocol.
In fact just to make it easy for you I’ll summarize what I’ve said so far:
– Linux is not free. Last I checked RH AS costs 2499. Pretty close to Win2k3 at 3999. You also get more years of support with Win2k3. (Included with that post was again that the linux-person as always was a moron.)
– Linux cannot power small, medium, or large size business for about 60 dollars a box. (Again pointing out the linux-user obviously was having a “help my IQ’s gone negative” moment.)
– Microsoft only wants money therefore they are bad!! So I reworded this as Publix, and every other grocery store is evil they only want your money – they don’t care if your malnourished!! (Again, unless linux powers the produce section – nothing bad about linux. Just linux people.)
So as you can see, I’ve gone straight to the source of the stupidity. The people posting here. Haven’t touched anything about how linux as a technology is. I think its fine. I just its not really getting anywhere because its users, are inherently MAJOR WHACKJOBS when it comes to making logical arguments about why the OS is good.
Btw, I hope the above about me not even owning an x86 compatable PC made you less ashamed to be a Windows XP user
Because seeing as I am an elitist (its part of the Mac-Fanboy status) I really don’t want you using Apples 
I think people are getting a bit bent out of shape here. The person really to blame is the author.
1) Okay, Microsoft does their own protocol better than Linux does Microsoft’s protocol. Does that mean Windows is faster? Of course not! Does it mean it might be a better solution for a mostly Microsoft shop? Quite possibly. But that’s something else entirely.
2) A couple of quick benchmarks run on an OS that has just recently been released do not a comprehensive study make. I’ve got a little bit of news for you. Speed, like security, is a reputation you build, not a breakthrough you have. If Windows 2003 is really what it’s cracked up to be, then great! It’ll let Microsoft build a reputation in the server industry for writing fast products. Right now, Windows servers (deservedly) have a reputation for being slower and insecure. If Microsoft’s changed that, then great. But you don’t change your reputation overnight. Look at it another way: Linux has a reputation for being hard to use. If tomorrow, RedHat came out with an OS that was as easy as BeOS, do you think people would immedietly change their mind and flock to Linux? Of course not! Change minds takes *time*. Time, effort, and persistence.
I find it rather interesting that they measured this at low load. Nobody really cares how a machine performs without a load on it. The steady-state of servers is to be heavily loaded. The main thing with Windows is not so much speed (in fact, for server uses, it’s been pretty competitive since NT 4.0) but performance under heavy load. It has very painful for Microsoft to switch Hotmail from FreeBSD to Win2K because the machines just couldn’t handle the load. I’d be really impressed if they’ve managed to change this.
I think you obviously are ignorant towards the Linux business model. Why are you taking a look at Redhat Linux Advanced server and Windows 2k3? Linux is a very scalable product, many major companies these days have adopted SuSE/Debian/Gentoo (you name it). All of which are very much low-cost solutions (free for download even) with good network-centric software management. The same goes for Redhat, you might as well just install off downloaded ISOs and pack yourself with server software. Support and documentation? Windows’s seriously lacks the documentation backend and support Linux recieves (arguably from commercial entities, and especially from the user communities).
Please do bring up better examples…
Regarding kernel-space HTTP, this is extremely bad (just imagine an off-by-one bug in kernel-space, bad). Linux’s “tux” and “khttpd” also offer this, but they are not as highly as adopted (which is good). I am looking forward to see how IIS6 ends up (an experiment or a success)….
>> – Linux is not free. Last I checked RH AS costs 2499.
Last time i checked Linux felt under the GPL. You can built your own RH AS with free products. dumb ass!
>>- Linux cannot power small, medium, or large size business
>> for about 60 dollars a box.
No even cheaper you are right. I use a PI 133 mhz thats is a
proxy,dns,firewall and dhcp server…
>>I really don’t want you using Apples
I hope you never open a xterm on your Apple, such a waste..
This is nothing but speculation and argumentum ad hominem. Until RedHat does actually sponser a similar test which shows that, you are just grasping at straws.
I think not.
Why can’t you just accept that it could be the case that Windows Server 2003 is faster than Linux? Why do you choose to rule out this possibility?
Oh quit the bollocks talking. Where did I rule out the possibility?
It’s natural that Microsoft having developed a product will want someone to do some indepedant tests to verify that what they believe about their product is true. It’s childish to assume that any company that takes a contract from Microsoft is so unprofessional that they would just make up the results. So instead of this silly nonsense, why don’t you either find a better source, a better methodology? Or at least provide some valid criticisms of the Veritest benchmark which prove it is totally and completely out of whack?
I’m guessing it is because there isn’t any such information that you know of, so you resort to slinging FUD straight back at Microsoft.
Oh please. I’m not disputing that Win2003 *may* be faster than Linux. What I am saying is that the article doesn’t present any concrete evidence to support this assertion.
Also, Microsoft sponsored benchmarks have been proven to be biased several times in the past. If you can’t find any evidence of this on the web, then I can’t help you – nobody can. As for FUD, you’re the only one spreading it. So shut up.
I said, small businesses and home users can run Linux and only have to pay $60 per box. A small business can upgrade their system for free, you don’t have to buy a new distribution every time you do that you know. A large business would be able to actively support research and development for open source products. Very large businesses are able to write off expenses, and citizens who have nothing to do with the choice, end up fitting the bill. The more support that Linux recieves the easier it will be for stakeholders to enhance open source projects. Open source has a lot of unrealized potential. It’s a model where people are in control of their platform and there is no reason to wait for a vendor to ‘innovate’. Vendors mostly just piece together open source projects, they work with some vendors for example RH and Oracle, but that’s only if you are going to use proprietary products.
Bottom line is that open source could use an object-oriented middleware platform like Java and .Net. That’s the only reason that I can fathom to support vendor products in this day and age especially for small and medium sized businesses. In the end I would rather support open implementation because it’s going to be more responsible to the users and more accessable. If more people pay vendors for everything, and to do everything for them, than vendors who take on the difficult task of innoviating will be trying to lock up their customer base and mislead customers, because the innovators hold the key. Innoviation is gradual, and open source should document research efforts, making the research responsable and accessable, gaining more manpower support, and leaving vendors to be more service oriented rather than the gatekeepers.
Why don’t they wait till the first round of Linux 2.6 distos are released then do a comparision. It seems very cheap that they’re comparing a 2 1/2 year old kernel against a brand new operating system. It would be like be comparing Linux 2.4 against NT 4.
As for the tests, the majority of servers are no SMP configurations and as for the NT4 customers, it is just been recently when SMP configuration server sales have picked up, up until then, the majority of sales of NT 4 servers were uniprocessor. If they wanted to do a true comparision, they would have done it on the typical NT4 seup, not the juiced up “2003” version of what NT 4 runs on.
As for speed, Linux 2.4 is still clunky on SMP configurations. Many parts of the kernel are still not fine grained enough to scale nicely. This should be corrected in 2.6, however, I wouldn’t hold my breath.
As for the alternative to SMB, I would suggest a NFS v4 + OpenLDAP combo and if you want security, pipe through OpenSSH/SSH. IIRC, one of the versions of Windows just recently have the ability to mount NFS shares along with accessing other UNIX resources.