Become a fan of Slashdot on Facebook

 


Forgot your password?
Close
typodupeerror
×
Unix Operating Systems Software

Kernel Musings: Unix and NT 357

Posted by justin++ from the damn-good-for-an-nt-mag dept.
Hulver writes "This is an interesting technical artical describing the differences and similarities between Windows NT and Unix. The article mentions Linux, but he is fairly dismissive, and very heavily biased towards NT being the best. If you are interested in how NT works and how it compares to unix, this is a good place to start. It might even open your eyes about how NT works. " I actually thought it was quite interesting, and only an eensy bit of it seemed inflammatory. Note that it's an article about the kernels of the operating systems, not the entire operating system.
This discussion has been archived. No new comments can be posted.

Kernel Musings: Unix and NT More

Kernel Musings: Unix and NT

Comments Filter:
  • I believe that the additional requirement for NT C2 certification was that the machine be disconnected from a network. Seriously.

    At the time, someone mentioned that we should get Linux B1 (or whatever) certified if disconnected from power mains.

    I wonder why all of the Top500 supercomputers in the world are non-NT machines? (Not even that many are Intel machines...of course, the first one is -- which counts for something)
  • AC Wrote:

    > NT SUCKS COCK

    Well, that must be one of the undocumented API's.
    It certinly sounds like an argument for going
    with NT.

  • Since when does AIX run on Alphas? Go away, troll.
  • That's what the article says. Really.
  • Hello anonymous coward,

    I hope you're joking. For your information, queso gives me this on your c2.org site:

    209.249.31.128:80 * FreeBSD, NetBSD, OpenBSD

    Hmmm, guess they don't trust NT that much.

    And could you provide and more specific link on that web site? I can't seem to find anything about NT + C2?

    I get it! You actually work for C2NET and you want to get /. right?

  • M$ (Score:1)

    by drwiii ( 434 )
    Hey NT must be good if M$ is going to impose incremental licensing [wired.com] on it, right? .. .. Right? .. .. ..

    .. .. ...

  • whats wrong with easy people?

    Being the source of all stupid and evil probably. People who want complex things to become easy often screw up those things for a lot of other people, that they haven't happened to think about.

  • Geeeez!
    You'd think that an OS that is as great as NT would be able to handle a little web traffic! ;-)
  • "NT's roots extend back to 1977 and Digital Equipment's release of VMS 1.0. Many core members of the future NT design team left Digital in 1988 to join Microsoft, which released NT's first version, Windows NT 3.1, in 1993. Thus, NT and UNIX have been evolving since the mid-1970s..."

    Ummm... that's a real leap. Just because some NT design team members also worked on VMS *does not* mean that 1)NT is descended from VMS 2) that NT has been evolving since the 70's. More FUD, sigh
  • With Unix's scripting capabilities, it shouldn't be too terribly hard to build a distribution disk that could set up an entire system with no manual intervention, and thus with no display at all, regular or serial. It would have to be pre-configured, of course...

    Another method that comes to mind is putting a network-capable kernel on the distribution disk and firing up a telnetd in the init scripts.
  • Umm, RMS is not a "Linux zealot freak." He applauds Linux's progress, but I'm quite sure he'll prefer the HURD kernel when it is at a usable level of completion.
  • Well, it would be nice, at least for bragging rights, if Linux implemented the necessary security features to earn a C2 rating. Sure, NT's C2 rating is fairly meaningless with all those requirements, but Linux can't earn a C2 rating even when it's not connected to a network and has no floppy drive, due to the lack of some required stuff like ACLs and auditing.

    If this does get done, it'll prevent the NT people from being able to say "we have a C2 rating and you don't!" Not to mention that some of the security features actually would be useful.
  • Linux can be installed without a videocard... perhaps the easiest way (off the top of my head) would be using the serial VT driver found in recent kernels.

    As far as SAMBA goes, he's very right about its performance; I saw some reviews after the last major release (whatever the heck it was) that were very impressive.
  • Posted by Nedwin_H_Longfellow:

    1.) the article brags on NT's scalability -- running on a max of 4 processors == scalability?

    HA! Even Irix, one of the worst commercial Unix variants available, runs on up to 256 processors, and probably more if the hardware would be available. Solaris has no problem scaling up to 64 or more in an E10, and probably into the hundreds, I just haven't seen it personally. Sure, Linux is way behind in this area with it's rather coarse-grained locking, but it's making very solid advances as we speak...

    2.) Security - typical M$ smoke and mirrors here. The only version of NT to ever earn C2 certification was a special version of 3.51, and it only earned the certification when the machine had no floppy drive, and one other criteria (forgot which...) You always see M$ bigots saying NT is C2, but notice how they skirt right over which version they're talking about?

    3.) They totally ignore reliability, which is everyone's main complaint about NT. There are some very serious design issues that I differ with in NT (such as putting video drivers in Ring 0, and allowing a poor video driver to crash an entire server, requiring that the entire machine reboot if you make one change in the network config, requiring you to keep reapplying service packs every time you load new software, etc.) but my main complaint against NT isn't the quality of the design, but the quality of the implementation. If M$ would spend even a 10th of their "research", advertising, or legal budget on improving the quality of their existing products, they would be quantum leaps ahead. I have never seen Solaris 2.5.1 or greater, HP/UX 10.2 or greater, or even Linux kernel panic without hardware problems (or in Linux's case unless I was dorking around with some kernel code :)

    Just my $0.02 worth, flame away if you love NT
  • Posted by jguest:

    Reminds me of Dan Quayle comparing himself to JFK.

    I was a VMS admin on a 7/24 hospital lab system. We had 98% uptime *INCLUDING* scheduled downtime and backups.

    I bailed as soon as I could when the hosp dedicated to an NT based lab system. Three years later, and they've finally decided that the app on NT can't hack it...
  • Posted by jguest:

    Come on folks, if they're going to state that NT is VMS with a GUI, then they need to also state that UNIX is MULTICS with portability.

    This puts the roots of UNIX back to the mid-60's -- significantly older than the VMS->NT pedigree.

    This reminds me of the Heinlein quote about meeting a small lizard who claimed to be a Brontosaurus on his mother's side....
  • Posted by Nedwin_H_Longfellow:

    You missed the point. 3.51 was the last version to be certified. What version are they pushing right now? Depending on whether you count vaporware, that would be either 4 or 5. How many million lines of code have been added or changed since then? Of course you'll probably tell me that multi gazillionaire Billy boy could get them to pass if he wanted to, he just doesn't want to waste the money, right? And nobody really needs a floppy drive or a network connection either, just like nobody will ever need more than 640K of RAM, right?

    I've heard that there are literally thousands of government offices installing plain old NT 4 because they expect it to be C2 secure. And from the drivel you read in these M$ rags, you would expect it to be too! I would love to see the people who make decent C2 OS's get together and start a class-action lawsuit against M$ for false advertising. (hint hint, anyone know any legal bloodsuckers working in the computer industry?)
  • Yeah, the best example I heard of was each user logging in to an NT box via WinFrame and triggering FastFind. Every time.

    Eventually the machine staggers to a halt under the load of a cheesy little app.

    ha ha!
  • NT desperately wants to be seen as 'Industrial Strength'. The way this discussion is framed leans heavily toward that.

    ie... NT is as old as Unix, and an evolution of VMS (a genuinely industrial strength OS).

    ie... Describing NT's architechture first, then going on to say that "Unix implements a similar approach", as though Unix were playing catch-up.

    ie... slagging Linux for not making kernel-level threads available, but ony casually mentioning that NT's kernel-level interface is undocumented and unavailable (the tech. may be beyond me here - is this a valid pont?)

    I think we're going to start seeing lots of articles that Take NT Seriously (TM), with no mention of stability, speed or overall appropriateness of this OS on a server. But simply taking that tone will start to leave an indelible impression on a "New generation of IT professionals".
  • Gee, must be running nt. We should collect some
    stats to compare how well nt boxes handle the slashdot effect as compared to Unix ones.
  • IMO This trys to clame a Unix legacy for NT that dose not exist.

    I have to agree. From the article:

    Many core members of the future NT design team left Digital in 1988 to join Microsoft, which released NT's first version, Windows NT 3.1, in 1993. Thus, NT and UNIX have been evolving since the mid-1970s, ...

    Claiming that NT is based on a Digital OS from the mid seventies because MS hired a few digital programmers is a huge leap. The author made it across the chasm, but he had to lighten the load and leave his credibility behind to do it. And claiming that both UNIX and NT date back to the mid seventies because of this is a further distortion of the truth; 1969 is _not_ the mid seventies. Lets go by ship dates: UNIX 1976; NT 1993. What we have here is a credibility gap of about 17 years.

    TedC

  • I never got to the article (for obvious reasons), but I wonder if the author left out the part about where NT's kernel realizes it has to do more than one thing at a time and decides to shut down all network resources, schedule all processes at real-time, flip the execution pointers to point to random instructions, and then start thrashing wildly as it completely falls over and fails to serve ANY CONTENT WHATSOEVER!
  • NT 4.0 does not, and is never going to have, a C2 level clearance.
    NT 3.51 (service-pack 3?) does.
    But only on isolated (non-networked) configurations.

    fwiw.

  • 5 years ago linux was just starting to be picked up for use in universities. 5 years later we see Linux exploding onto the scene. Not a coincidence.
  • Is when he writes a couple paragraphs on "Which OS is Better?" where there's some good arguments for UNIX, and then the next paragraph is "Which OS is REALLY better?" What's especially funny is in the previous part he says the only way to really say which is better is by using standard benchmarking utilities. But to find out which is REALLY better you just have to know NT is the choice of a new generation of IT professionals..

    I'm still chuckling about that..
  • You could use Samba to export the directory instead of NFS - that would work in much the same way (although without ACLs).

    I do agree with the point to some extent - although I think it is one of the areas to cause the largest headaches on NT as well. Ever tried dealing with permissions problems on a server for NT? Nightmare.
  • Look at the "NT News Analysis" ( http://www.winntmag.com/Magazine/Article.cfm?Issue ID=97&ArticleID=4651 ) instead.

    "According to intellectual property lawyers, the Linux licensing agreement binds any developers who produce software using components of the Linux OS (e.g., libraries, runtimes) to release the source code for their additions (i.e., applications) to the public domain."

    /mill
  • NT architecture is definitely more modern than UNIX, incorporating many desirable features (and some undesirables, as many of you have been expounding here for long time).

    You're right.

    however, nt seems to be the poster child (poster devil?) for E raymond's theory about how "cathedral" development model fails in reliability. ms recruited many many high quality talents with their market clouts, yet they still put out low quality products (at least in terms of reliability).

    Actually, read "Rapid Development" (MS Press). Don't get too hung up on its advice, although by and large it's a very good book.

    Anyhow, the author points out that MS did the opposite of this -- they didn't pay money to get the best, they instead used an internal hype campaign to get the excited programmers eager to work on the project. There's a saying about that -- there are old programmers, and there are bold programmers. But there are no old, bold programmers.

    So MS traded enthusiasm for writing "the best OS ever" (their words) for experience and competance.

    The result is self-evident.

    Will bazaar-style open source ("open development", to coin a term) make it possible to be both experienced, competant, AND excited? Time will tell. I hope so.

    -Billy

  • Cute. Your little saying is not new. It originates in the same phrase with airplane pilots.

    Yup. Of course! The copyright's expired :-).

    However, it implies that bold pilots end up dead, not that old pilots are generally useless, like one would infer from your version.

    Read again -- like you say, I'm quoting the saying almost verbatim, and my message was otherwise extolling the virtues of experienced programmers. Don't know how you got that interpretation.

    -Billy

  • I have a 486DX-66 with 16MB of RAM, which
    boots Linux and runs Apache just fine.
    It's currently serving up to 1000 users a day,
    without pushing the load average over 0.05...

    Danny
  • IF everyone thinks NT is better, guess what! we all lose. All the innovation with linux will stop because gnu will turn into a windows thing crippled by lack of api's (thanks to microsofts deliberate attempt to break the public ones and hide the private). Come Guys no one ever gets fired for buying microsoft software. MY job and your will be gone or you can still have it if you only use ms wrpducts for everything. Oh! Guess what! Your job will be gone! why do you need to to hire consultants when you can just visit microsofts web site and buy the package for your needs. Why do you need a consultant to find out which programing suite you need or which database you need when you or any idiot can just order SQL server or Visual studio. Microsoft has a solution for every problem right!:-)

    I DONT WANT TO PAY $1300 FOR VISUAL C++ AND ANOTHER 700 FOR VISUAL BASIC! MICROSOFT CRIPPLED ALL THE CHEAPER VERSOINS OF VC BY MAKING THE PROGRAMS RUN SLOWER ! ITS PRATICALLY THE SAME PROGRAM BUT WITH A CRIPPLED COMPILER! I HAVE NO CHOICE BUT TO USE VC BECAUSE ONLY VC CAN USE THE API"S OF WINDOWS! WHO THEY HELL DO THEY THINK THE ARE TO TELL ME WHAT TO BUY! I think its awefull that I am closed from the real (windows) world. I bought caldera linux out fo rage because I wanted to do some programming. I am not a unix tech guy but just a hobbyist who refuses to reward Gates for his crippling of the api's to visual studio only! I want to write high performance grahics apps with directx and opengl and microsoft has totally used my own computer agaisnt me. Why do I need to be punished by ms by not reading and sharing info with office just because I chose to buy my own os. Microsoft has ilegally and unethically abuased its power and we linux users suffer! WOuldn't it be great if office 2000 didn't have encyption technology in its .doc and .xls files so you can share data with the world or it would be greater if everyone used an open standard like html or any other txt format, so it wouldn't matter what wordprocessor you used. Everything was open before Billy Boy and Big BLue came along and changed everything. As you can tell I have a big type A personality and I can invision BIlly grining (just as he did when he told the press that Ie was part of windows)when he sees me fustrated with windows. I am tired of microsoft insulting mine and all other it profesionals intelligent's. I bet all the employees were laughing their asses off when they heard Muth bash linux. BIll just thinks he can do whatever he wants and can make all of the IT professionals think whatever he wants us to think. ITS NOT UP TO MICROSOFT TO SET STANDARDS OR TELL MY BOSS WHAT TO BUY> Its their job to try to convince us but not to call my boss directly and tell him to fire me (MS DID CALL MY BOSS TO TRY TO CONVINCE HIM THAT NT IS THE ONLY SOLUTION AND THAT GUY WHO ISNTALLED LINUX (me) SHOULD BE FIRED BECAUSE I WAS INCOMPETANT FOR NOT SELECTING NT!). I am not MAKING THIS UP! My boss never called ms but rather ms called us and threatened us. They are like the mobfia. These bull performance charts are just one small piece of the pie. I hope the DOJ SLAMS THEIR ASS!

    aaah the previous poster is right. This does feel alot better.

    oh and one more thing. BIll Gates on Germany radio stated that windows 98 doesn't have any bugs and that if it crahses THEN ITS YOUR FAULT! Now. How does this make all you support people feel?

  • Not for long. At approximately 13:45 UTC Sunday March 7, the server is down again.
  • Hmm, I dunno - what do you make of this: When my buddy is logged in at the console he gets his drive letter mappings (say under his login drive "S:\" is mapped to a particular directory on another machine). If (while he is logged in) I telnet in and log in as myself, I'm still stuck with his drive letter mappings not mine. I'm not trying to be a wise guy, but this is something I've dealt with within the last week so it came to mind. Comments?
  • Actually, I would think that brk() still needs to be extended to support 64 bit on 32 bit platforms, so there may be others, as well.

    And I haven't checked, but are you sure that glibc 2.1 supports 64 bit mallocs? How could it, if brk() itself doesn't?

  • ahh.. yes.. i can finally go to sleep now.. thank you.. that felt good..

  • I believe that is what the article said about half way through. NT is C2 certified on a non-networked Standalone system. Duh..

    Sure you can reboot a unix machine into runlevel 1 but what good is a secure standalone?

    Idiots....



  • Oops!

    Oh sorry... wrong forum.

    ;-)

  • On Solaris I can set my files' rwx permissions for everybody, but since I can't create or modify groups I can't have any finer grained control.

    Yes, you can, if you're running Solaris 2.5 or later - if the file system is local or on an Solaris 2.5 or later NFS server, set an ACL on the file (as another poster noted, man setfacl, although I've not found a man page that explains well the semantics of Solaris ACLs (they seem to closely resemble the ACLs in the POSIX 1003.1e D15 draft).

    If I had Linux on my laptop I'd have to NFS export the directory.

    That depends on the machines to which you're explorting it. If the clients support SMB (Microsoft-flavored OSes do, as does Linux), and if you have Samba on your machine, you could SMB-export it (as you can with NT) with Samba.

  • In theory, a microkernel design is more portable than a monolithic kernel.

    Perhaps Microsoft should have considered using such a design, then. I've seen nothing to indicate that NT is anything like what I generally hear referred to as a "microkernel" - file systems, device drivers, networking stacks, and even low-level graphics in 4.0 run in kernel mode (I suppose a desperate Microsoft marketoon could try distinguishing between the NT "kernel" and the NT "executive", but somebody could partition the routines in a modern UNIX-flavored kernel in the same fashion), invoked via procedure calls, rather than in server processes.

    Yes, some parts of the Win32 API are implemented in the Win32 subsystem process, but if that qualifies it as a microkernel, one could argue that an automounter makes UNIX a microkernel....

  • I think the little paper-flying things are in the kernel alredy.

    I've not seen anything to indicate that they are. Stuff at, I have the impression, the level of the drawing routines in Xlib may be in the kernel, but stuff at the toolkit level lives in userland, as well as stuff implemented atop it.

  • www.interix.com

    Assuming that citing a domain name that gets you to Softway's pages about Interix is an attempt to defend the assertion that NT is a microkernel, I don't consider the ability to have multiple environments for different types of programs to be sufficient reason to dub something a microkernel - were the Wine folk to add extra system calls to the Linux kernel to assist it, would that render the Linux kernel a microkernel?

    In addition, how much of Interix is implemented in the Interix subsystem process and how much is implemented in libraries called by programs running under Interix (with those libraries perhaps making NT system calls)?

  • Actually, 3.51 was never certified. 3.50 was, and I know that the following 2 conditions applied to the certification:

    1) No floppy drive.
    2) No network connectivity of any kind.

    Kinda sad how they have been getting away with claims of C-2 certification for all these years.
  • I would think it can't be very hard to replicate the NT login screen and make a fake program that can steal passwords.

    Unless it is impossible for a program to stop the Ctrl+Alt+Delete? I suppose that would make a real clone impossible. However a tiny hardware modification (probably doable with only access to the keyboard) would make this pointless.
  • Then they realised that NT is useless if video doesn't work 95% of the time, so they moved it to the Kernel to eke out a little bit more speed.

    Oh, so is THAT why my NTW is BSODing on me twice a day?

  • In the paragraph where he discusses the portability of various Unix variants, there is no mentioning of Linux. In the spec-ratings, not either.

    Also, he tries very much to keep things theoretical. Nowhere anything about uptimes or reboot issues (for quite a lot of NT setting changes you need to reboot). Practical things count!

    Furthermore, most of those 'hobbyists' happen to be IT students who are very very likely to become 'a new generation of IT professionals' in a few months/years.

    And indeed Linux machines _are_ being used in commercial production environments. I for one get paid for installing Linux for companies, as file/mail/dial-in servers. Not ready for the environment? Ever so ready!
  • if I remember correctly that 'C2' NT 3.51 had to be stand-alone. Why didn't they just require it to be turned off?!
  • You have a point about the 'kernel uptime', although when both kernel and tools come from the same manufacturer, MS in this case, and uptimes aren't good, questions arise about implementation of both tools _and_ kernel. But other people have said more knowledgeable things about design, which is said to be good, and implementation, which is said to be not so good, regarding the NT kernel.

    Furthermore, I don't remember flaming anyone.
  • I use both NT and Solaris at work and there is something to be said for NT's file sharing.

    On Solaris I can set my files' rwx permissions for everybody, but since I can't create or modify groups I can't have any finer grained control. I either grant a permission to everybody or nobody. If our company was more geared towards Unix support I suppose I could make a phone call to an adminstrator and get a group created or modified eventually.

    If I want to share a file or directory on NT I can make a list of people who I want to have access. I can give read access to some people and write access to others on the same file. I don't have to have any special privledges, I just right click on any file or directory and select peoples' IDs and grant them each whatever permissions I want them to have.

    Furthermore, NT's file sharing is based on *people*, not *machines*. My Solaris machine is on the network full time and my home directory is on a server. I can share files because other people have access to the same server. This wouldn't work for a laptop. If I want to share a directory on my NT laptop I just right click on it and turn on sharing. Whenever I'm on the network the shared directory will be visible to the specific people I granted permission.

    If I had Linux on my laptop I'd have to NFS export the directory. That would require knowing what machines are likely to be used by the people I want to share files with. On NT if I grant "fred" access to one of my directories, "fred" can access it from any NT machine on the network as long as he logs in a "fred". If "fred" normally logs in to a machine called "tophat" I could export an nfs mount to "tophat", but if "tophat" died one day and "fred" went over and logged into someone else's machine he wouldn't be able to nfs mount my directory from that machine. He'd have to call me and ask me to add that machine to the nfs exports (which I couldn't do even if I were in the office instead of at an all day off site meeting, I'd have to call an administrator).
  • -- there are old programmers, and there are bold programmers. But there are no old, bold programmers. --

    Cute. Your little saying is not new. It originates in the same phrase with airplane pilots. s/programmers/pilots/. However, it implies that bold pilots end up dead, not that old pilots are generally useless, like one would infer from your version.

    -kabloie
  • But what happens when it crashes? YeeeOUCH!
  • I laughed my arse off. Then I realized he was serious.

    NT is here to stay and it's becoming the choice of a new generation of IT professional.

    I refuse to comment. Despite the fact that he himself admits that NT does not perform anywhere near UNIX, his opinion is that NT is competitive with UNIX on high-end servers.

    He also refuses to acknowledge the "homegrown Unix variant" Linux, stating that "a large percentage of the Linux installations are still in the realm of the computer hobbyist". Proof of this, Mr. Russinovich? To be sure, there are many of us who have adopted Linux as our home OS and continue to happily hack away at it, but signs have shown nothing but continued acceptance of Linux in the business market.

    All in all, it's standard FUD, exactly what one would expect from a source like www.winntmag.com.

    ~Dan
  • To: webmaster@winntmag.com
    From: jamie@mccarthy.org
    Subject: Unable to reach article

    Dear webmaster,

    I read about your NT-Linux comparison on slashdot.org. However, I was unable to get through to your Microsoft-IIS/4.0 website because:

    Error Occurred While Processing Request

    Error Diagnostic Information

    Server busy or unable to fulfill request. The server is unable to fulfill your request due to extremely high traffic or an unexpected internal error. Please attempt your request again (if you are repeatedly unsuccessful you should notify the site administrator). (Location Code: 25)

    Please inform the site administrator that this error has occurred (be sure to include the contents of this page in your message to the administrator).

    Good luck getting it back online. I feel your pain...my company used to run its webserver on NT too...

    Jamie McCarthy

  • Why not? While it may not be found in the dictionary, it's a lot more likely to be understood than many words in the dictionary. And isn't that what writing is about, being understood?
  • Ain't Windown NT GREAT?
  • The author complains about cost per transaction being much higher with UNIX than NT, and yet he's completely dismissive of UNIX running on commodity hardware! (Linux/BSD)

    Apples and Oranges! If the NT benchmarks had been done on something exotic like Alpha hardware, or PPC hardware the numbers would be quite different. Everyone knows the "mainstream" UNIX vendors are raping their customers with price, but to exploit that in a benchmark is silliness.

  • I've only been doing this stuff 3 years(just barely)so a lot of that article (when I could finally get it to load :-)was way over my head (All those years of Radio-Electronics and {the original}Popular Electronics, I devoured the analog and HI-FI stuff and just sort of skimmed the digital unless I had to fix something with TTL chips in it).
    So let me see if I got this straight.
    Linux is mostly GNU, GNU's not UNIX, and Linux is just another form of UNIX.
    NT's all original innovation, no "borrowed" Digital code in it, but it traces it's roots back to when the guys who wrote it were on Digital's payroll. (Maybe they were writing NT after hours on their own time before jumping ship.)
    Clear as FUD!

    P.S. Katz on C-SPAN, Monday (3/8/99), talking 'bout his book! Is there a Slashdot Effect for TV ratings?
    (Yeah, but he's OUR gasbag!:-)

  • Bais (Score:1)

    by Jerry ( 6400 )
    I was very active selling Apples and writing Apple BASIC apps to run on them in the early 80's. My memory of XENIX was that it appeared as the OS on a model or two of Radio Shacks's computers, but never made a big dent. I think this is the "suprise" the author refers to when he says that XENIX had the largest "market share" of UNIXies at the time. As I also remember, that market share was largely home and very small business. So, M$ gets praise for having the "largest" market share of UNIX computers in the early 80s even though they were personal-home computers, and Linux gets dissed because a large part of it's market share is personal-home computers.
    How do you say "double standard" children?
    At the government agency where I program, my self and many others installed NT 4.0 on our workstation when it first came out. As the months passed we reverted back to Win95 because of problems relating to speed, stability and NT's "user friendlyness" forcing us to do things in ways we didn't want to do them, I guess to protect us from ourselves. "User friendly" gets in the way when it assumes that every user is an idiot who has to have their hand held (or slapped) all the time.
    A couple of weeks ago we installed our first Linux server, replacing the NT that had held our Oracle 8 database. Results: an immediate four-fold increase in speed and the nearly daily reboots have ceased.
    When app development issues (which language, which toolkits, which Xclient, which VCS) get resolved by the suits it is my belief that 30+ servers and 300+ workstations are going to get a drastic overhaul. All from one CD. The economics and the speed and stability are going to force it.
  • That only proves that NT executed the BABBIT API call without warning! That's why you can't trust it - it's so unstable.
  • The Graphics wasn't in the kernel to start with because most video drivers are written by the manufacturer not MS.

    Then they realised that NT is useless if video doesn't work 95% of the time, so they moved it to the Kernel to eke out a little bit more speed.

    Chris.
  • If this is the same article, I quoted it on http://www.windowssucks.com

  • /.ed (Score:1)

    by fatboy ( 6851 )
    Nope, I restarted my webserver to do rev lookups.
    Try it now :)


  • OK, its not a mirror --- I couldnt get to the article so I didnt know which one it was. Therefore you are correct, I am scum.

  • <joke>
    I checked out this guy's claims, and they're true! I put each of the following OSes onto the 16 DEC alphas I have in my living room, and did some web performance testing by refreshing a page in my browser. Here's what I found:

    Windows NT running IIS 4: 15 pages per minute (until it crashed)
    Windows 95 running PWS: 20 pages per minute
    RedHat Linux/Sparc running fdisk: 0 pages per minute
    AIX running SSH: 0 pages per minute
    MacOS X Server running MacAmp: 0 pages per minute (and they said it would be good!)
    PalmOS 3 running Memo Pad: 0 pages per minute

    So as you can quite obviously see, Windows NT is a far superior operating system!
    </joke>

    --
  • OK, I have used NT. In fact, using NT is what inspired me to learn about Linux. After reading this article, though, I come away with the clear impression that NT has a much better design for its operating system. Some things, in fact, that I would say are better in NT. But my experience dictates otherwise.

    So, if someone could explain to me:

    a) What are (seriously) the things that NT does better than Unix/Linux? (Besides BSOD)
    b)Are those things being developed?
    c) If NT is "better" in some ways, why does it suck so badly?
  • NT 4.0 does not, and is never going to have, a C2 level clearance.

    NT 3.51 (service-pack 3?) does. But the person who got that clearance, Ed Curry, has publically stated that the changes to NT 4.0 have weakened security so much that it cannot meet the standard.

    Doesn't stop Microsoft from spreading their lies though...

    Regards,
    Ben Tilly
  • I work for what some would call a shark-tank (I'm a government contractor).

    At one of my contracts, we went in to replace NT boxes running simple web services (we were replacing them with Sparcs running Sol 2.51)

    It would be too generous for me to say that the NT system was too unstable. Those poor bastards who set up the system had to install a watcher. It would ping all of their NT boxes (about two dozen) until one of them failed. The watcher program activated a modem, and dialed out to a pager when one of the systems failed.

    Then one of the poor bastards would come scurrying down into the server room -- and reboot the NT box. Those dudes would be paged twice a day. I know, I sat at the desk next to the modem. I held back a huge belly laugh each time the modem dialed out.

    I feel sorry for you NT people, I really do.

    NT zealots: You won't be taken seriously by the Unix/Linux crowd until your OS stops crashing once a day.

    The winntmag.com site should think about investing in that watcher program -- or better yet -- install a reliable OS for the webserver.

  • C2 is an extremely ugly business. I personally would not volunteer to work on any machines that are C2 level compilant -- cause you can't do anything on them. I've unfortunately had the most distasteful duty of making a proprietary UNIX operating system C2 compliant. It was darned hard to do that -AND- have the system do anything useful.

    Linux would probably be C2 level if (and not to bust on Linux -- I like it a lot, but C2 is ridiculous):

    1) The code was proprietary, and evil hacker scum couldn't look at the code because the best security is by obscurity.
    2) The permissions on all files were set to 700
    3) Shadow passwords were enabled
    4) All files were owned by root
    5) No floppy drive
    6) No Internet connection
    7) No Network card
    8) No third-party "hackerware"
    9) turn the computer off
    10) solder the power switch into the off position.

    Next time somebody brags about their system being "C2" compliant, think "Ah, you poor dumb bastard"
  • My favorite paragraph of the article follows. It would be nice if the author knew the correct name of a certain Sun CPU, knew that AIX runs on PowerPC, RS/6000, and Power2 architectures, and might have known that "codeveloped" is not a word in the English language. It would have been nice if he realized that Windows NT used to run on MIPS architectures and had a short-lived PowerPC version as well. Still, the comparisons presented in the article are well-researched.

    It sometimes seems as if NT gets less portable by the day. NT currently supports the Alpha and x86 architectures. Although you can probably find a version of UNIX that runs on any given hardware platform, the leading commercial UNIX releases are even less portable that NT, running only on their vendors' proprietary CPU type, and sometimes on the x86 as well. For example, Sun Microsystems developed Solaris for the SUN Sparq chip but ports Solaris to the x86. IBM's AIX runs only on the PowerPC chip, which IBM codeveloped with Motorola.
    Kris

    Kriston J. Rehberg
    http://kriston.net/ [kriston.net]

  • If you think you got the balls to insult everyone who reads slashdot: "looks like the typical /. iq remains where it's always been - below that of the intelligent general public.", you might want to tell us who you are, tough guy. Just a little something special from you friend calx, who isn't even able to look at the article, because their server is down.
  • >If I had Linux on my laptop I'd have to NFS export >the directory. That would require knowing what
    >machines are likely to be used by the people I
    >want to share files with. On NT if I grant "fred"
    >access to one of my directories, "fred" can access
    >it from any NT machine on the network as long as
    >he logs in a "fred". If "fred" normally logs in to
    >a machine called "tophat" I could export an nfs
    >mount to "tophat", but if "tophat" died one day
    >and "fred" went over and logged into someone
    >else's machine he wouldn't be able to nfs mount my
    >directory from that machine. He'd have to call me
    >and ask me to add that machine to the nfs exports
    >(which I couldn't do even if I were in the office
    >instead of at an all day off site meeting, I'd
    >have to call an administrator).

    And this is a GOOD thing?

    Hows about I just walk into your office with a new laptop with a "fred" user account on it and login to your machine.

    I'd much rather apply a network-address based control than a login-based one.

    It seems that this type of security model is based more on convenience than anything else...
  • Hmm, about 10 months ago, I was one of those linux using students. Then I graduated and went to work for a place that had NT on all the desktops. (Most of the development actually happens on the solaris servers, because we use clear-case, and the compilers we use are available for solaris.)

    After about a month I was using linux on my desktop, occasionally rebooting to NT to use word/excel. Now, there are at least 3 linux servers at work, 2 of which I set up, and we have several more in the works. (One of those linux servers in the work will replace our last NT server.) Also, several of my co-workers have switched to linux on their desktop, and there are some others who are playing with linux at home.

    So, if every one of the current linux using students graduate and have this sort of impact, M$ has really got something to worry about!

  • 2.) Security - typical M$ smoke and mirrors here. The only version of NT to ever earn C2 certification was a special version of 3.51, and it only earned the certification when the machine had no floppy drive, and one other criteria (forgot which...) You always see M$ bigots saying NT is C2, but notice how they skirt right over which version they're talking about?


    The "other criteria" is that it can't be attached to a network... I'm not fully conversant with C2 specs, but IIRC that means not having a NIC installed, not just not being plugged in. There's a secure server OS !
  • No, unix zealots are too busy whining about Microsoft and NT to discuss any of the benefits of unix.
  • ... And slashdot.org isn't propaganda favoring linux???
  • More likely GDI was moved into kernel space when they realized people were able to hook all GDI calls and translate them to X protocol messages and you could use NT from, *gasp*, a non-windows workstation. Without even paying mucho denero and signing NDA's for source licenses.

    Couldn't have that, no sir.
  • Everything may be a file in unix, and unix may have ACL's, but it seems neer the twain shall meet in some cases.

    Try this on solaris:
    setfacl -m dude:rwx /dev/cua/a

    You lose. What burns me about NT is that it could expose every internal structure as a file, ala a /proc filesystem. It just refuses to, treating you as an idiot who's just not "ready" for that kind of information.
  • It's a new processor from Sunni Microsystems. Everything else is shiite in comparison.

    Man, I'm gonna pay for that one in the next life.
  • >>IN NT YOU HAVE TO BUY(!) SOFTWARE TO HAVE >>TELENT..
    >Wrong. MS has a free UNIX connectivity pack with a telnet daemon.

    The one that crashes every time someone disconnects? The one that has dire warnings of unstable beta software written all over it? The one that is unsupported and doesn't come with NT, meaning I can't telnet into a newly installed box?

    What a joke. This is TELNET. God forbid MS should try something like ssh.
  • I'll let you in on a little secret:

    man setfacl

    I didn't know about that til I came to work for Sun. Enjoy. :) I agree NFS sucks though. So does CIFS. Just in different ways.

    For me, there's one major reason Unix FS's just blows the doors off of NT's: symbolic links. I hear NT5 will have those. I hear NT5 might even exist. I hear voices too.
  • My favorite is this:
    "Internal data structures limit NT to using a maximum of 32 processors, but licensing limitations usually restrict the number of processors to 8 or fewer."

    HAHAHA
    I just don't understand why you would want to use this operating system.
  • Both benchmarks he uses are wrong. One measures the performance of web serves, the other the performance and cost-benefit of databases. The Web test he uses was also used by ZDNet to "show" that M$ Web Servers were faster than Apache (the non-production NT version of Apache...).

    In the database test case he is lying deep. He is quoting from the transaction test. SQLServer does not even appear in the top-ten performance list. But gets all ten positions in the cost-benefit list (does anyboy else sees dumping here?) with very poor performances. This test only shows that MS SQL Server is a expensive alternative to MS Access. He obviously refrain from quoting the warehouse test (large data sets, low transaction rates). SQL Server is not even listed.

    Even ZDNet published file-serving and print-serving tests showing Linux beating the crap out of NT (both running in exactly the same hardware).

  • I heard from more than a few people that say gcc is crap really and egcs is the way to go, perhaps anything that starts with G should be dumped, with out mentioning anything.

    `I heard...'--beautiful; I love people who can't think on their own....
    What's egcs made out of?
  • Of course all software doesn't have to be free, but at least have some respect for someone without whom there would be no Linux, because there would have been no free compiler to write it, no free debugger to fix it, no free shell to boot in, no free shell commands to use with.

    There'd be no operating system to host Torvalds' kernel--that is why there'd be no Linux.

    As others have said, kernels don't do anything by themselves.
  • The site's down. Do I assume that the slashdot effect got it.. or does it have to be publicaly announced to believe that...
  • Yep! I managed to catch the article on slashdot right hot. The thing was still working there. After almost one hour I lost contact with NTMag.
    SLASHDOT EFFECT starts to possess nuclear properties...
  • More. In a very, very broken comp I installed NT. The thing crashed every hour. Sometimes it was crash time. In the same box Linux managed to freeze twice in a month! And I may tell you that I didn't use it just for typing ls -l. The little thing was app fried several times. It even ran with Quake2 (SVGALib) + X + Netscape (downloading compressed files) + gcc (compiling kernel). It cried like Hell. But it kept working.
    Besides on a IBM's NetFinity with NT Server I wonder until now why nwadmin32 + netscape ONLY, managed to eat up all memory (reinstalling everything didn't help).
  • I confess I ran through the article in a fast way. I hoped to read it more calmly at home but it seems that "Slashdot Effect" killed NTMag.

    However the very first reaction I had, was that this guy was trying to demonstrate that NT is a UNIX like anyone else. Funny but that was my very first reaction on looking through the whole article at fast speed. It seems that M$% is changing the politics of a "better OS than UNIX" to a "better UNIX than UNIX" :)
  • Yes, and when you MS Lackies stop posting as ACs pigs will fly right?

    I really wish /. went the Drudge route - post webserver logs. It's always good to see how much of his trafic is whitehouse.gov.

    I'd love to see how much microsoft.com traffic we get here. Seeing as they tried to buy linux.com ( to either sit on it or use it to spout FUD )
    --
    James Michael Keller
  • Heck yea!!! Hook a good fast NT server under my desk and I'll work all day! ;)
  • Well, where do we throw Mach in at?
  • Imagine this scenario: you run a Win32 console app from the telnet session, expecting it to never interact with the GUI. The app is stored on a remote file server, accessed via a drive mapping. NT attempts to demand load executable pages from the binary, but encounters network congestion, and times-out on the packet. Poof, NT kills your console app, and displays a dialog message box. Where does that dialog appear??? On the frigging graphical console. Can you resume use of your telnet session? Most likely not. Gotta wait for someone at the graphical console to hit the 'ok' button on the dialog.

    But NT does offer mechanisms for simultaneous graphical desktops on the same machine. Just use CreateDesktop(), and SwitchDesktop(). They can be created under the auspices of other users, using NT's impersonation routines. But you then encounter other multiuser limitations of NT, such as the redirector! Once the redirector creates a drive mapping, it becomes globally accessible to all active users of the machine, regardless of their privileges.
  • Imagine this scenario: you run a Win32 console app from the telnet session, expecting it to never interact with the GUI. The app is stored on a remote file server, accessed via a drive mapping. NT attempts to demand load executable pages from the binary, but encounters network congestion, and times-out on the packet. Poof, NT kills your console app, and displays a dialog message box. Where does that dialog appear??? On the frigging graphical console. Can you resume use of your telnet session? Most likely not. Gotta wait for someone at the graphical console to hit the 'ok' button on the dialog.

    But NT does offer mechanisms for simultaneous graphical desktops on the same machine. Just use CreateDesktop(), and SwitchDesktop(). They can be created under the auspices of other users, using NT's impersonation routines. But you then encounter other multiuser limitations of NT, such as the redirector! Once the redirector creates a drive mapping, it becomes globally accessible to all active users of the machine, regardless of their privileges.
  • You don't need all that much money to have a system that blows the most powerful NT system out of the water. Just look at what BSD does on a daily basis for ftp.cdrom.com and compare that to what NT can do. NT isn't ready for the internet.
  • Hows about I just walk into your office with a new laptop with a "fred" user account on it and login to your machine.

    Uh, you'd have to know fred's password. It's no less secure than having two 'fred' accounts on two different Unix boxes.

  • I think what we're seeing here is a certain Linux elitism -- "Linux doesn't have ACLs, so there not really useful." Unfortunately this is counter to the experience of thousands of production NetWare and NT administrators.

    Seeing an ACL like this is quite common on any NT or NetWare file sharing box:

    Report Readers - Read
    Report Editors - Read/Write
    Administrators - Full Control

    Yes, I know that there are workarounds for the brain-dead Unix permissions, but please let all of us unenlightened system administrators know how the above ACL is not useful "in practice".

    --
  • To follow myself up --

    (1) I fully expect that someone will develop "ext3" or whatever that has ACLs and other desirable features within a year or two.

    (2) I actually do know the difference between "they're" and "there".
    --
  • "Full Control" means you can change the ACL, aka you are "owner". Why not make the user's owner? Because some of them will grant "Full control" to "All Users" for what ever reason, and then come back and bitch that there's a security hole -- believe me, it happens.

    Your example breaks, because "other" is everyone else, not just "Report Readers". Why expose information to people who are not supposed to have it? Furthermore in an NT domain or an NDS tree, "Other" could be 100,000 or more people that you've got no control over, not just your 8 local users or whatever.

    As other's have pointed out, in a large distributed system (like AFS), you need ACLs.

    As for Applications which bypass normal OS security, take a look at Lotus Domino -- it has an even more complex ACL model than NT or NetWare does.
    --
  • Funny, I was just starting a GTK to OS/2 PM port and noticed this in the win32 mods to glib:
    [From memmory]
    PeekMessage( MGG *, HWND, NULL, 0, 0, PM_NOREMOVE);

    The OS/2 PM Eqiv of:
    WinPeekMsg( HAB, QMSG *, HWND, NULL, 0, 0, PM_NOREMOVE );

    I guess that somebody decided that the PM_NOREMOVE must be PeekMessage not Presentation Manager so they didn't have to change the constant.

    Sorry, but I found it hilarious.

    If anybody else is working on a PM port of GTK I'd be interrested in working with them.

    shaun@tancheff.com

  • WTF: This 486 w/16M needs an upgrade, maybe you can /. it
    http://shaun.tancheff.com/NtMag.html [tancheff.com]

Slashdot Top Deals

Work is the crab grass in the lawn of life. -- Schulz

Close