Follow Slashdot blog updates by subscribing to our blog RSS feed

 


Forgot your password?
Close
typodupeerror
×
News

Time To Re-Evaluate Microsoft's Linux Myths Page? 292

Posted by Cliff from the do-these-still-hold-true-today dept.
cluge asks: "MS still proudly presents their Linux Myths Page and recent Dell commercials show Linux as a 'lower end' solution. This seems even stranger coming from Dell when you look at all the the Linux solutions that they are offering. The comparison made in their print adds compare 2 machines with vastly different amounts of memory (or so has been reported here and elsewhere). With a new kernel coming out, should the Linux community tune a couple of machines and set up an open test? Test the following: static Web serving, file sharing (Samba or NTFS), and routing performance. Having Linux knowledgeable people run the test allows performance tuning for the application being tested. The testers comments and recommendation on performance tuning will be valuable to the entire Linux community and the tests will let the community know where it stands, win, lose or draw." So how much difference does a year make? Do any of the claims on that page still hold any merit today?
This discussion has been archived. No new comments can be posted.

Time to Re-Evaluate Microsoft's Linux Myths Page? More

Time to Re-Evaluate Microsoft's Linux Myths Page?

Comments Filter:

  • Dell & Linux (Score:1)

    by Anonymous Coward
    This seems even stranger coming from Dell ...

    Indeed. I believe Dell Corporate speak with forked tongue.

    Not long ago my company bought a Dell PowerEdge box to run Linux and the T.REX firewall software on. When I ran into problems with the tape drive, Dell support noted that "It must be a problem with the drive itself. If you were running NT, we'd suspect the OS, but Linux and the Linux SCSI tape driver are solid as a rock." Or words to that effect. (It was the drive, btw.) When I questioned that tech, and another, on the issue: both assured me that I had interpreted those comments correctly. I got the impression, perhaps mistakenly, that Dell tech support was generally having less problems with Linux than they were with NT.

  • He has proven himself completely igorant of how ma and pop ISPs run websites. Many cheap sites use big rooms full of cast-off Pentium I machines.

    Unfortunately for VA Linux, the era of the screwdriver shop server is coming to an end as th Net turns into more than a string of Lemonaide Stands.

  • Re:Never mind 99.9, try 99.999 (Score:1)

    by Anonymous Coward
    AFAIK, there is no good ssh1/2 & scp client/server package for 95/98/NT/2000.

    For a client, I use PuTTY. Does telnet, ssh, scp. Good ANSI support, scroll buffer, very light and fast. Recently became legal in the USA. Good stuff.

  • Frankly, any application that you are considering using NT or linux for is probably going to involve some scheduled downtime that negates any uptime stats.

    Its a nice bragging point, but in the case of low-end servers, its not a big deal.

  • Linux performance and scalability is architecturally limited in the 2.2 Kernel. Linux only supports 2 gigabytes (GB) of RAM on the x86 architecture,1 compared to 4 GB for Windows NT 4.0. The largest file size Linux supports is 2 GB versus 16 terabytes (TB) for Windows NT 4.0. The Linux SWAP file is limited to 128 MB. In addition, Linux does not support many of the modern operating system features that Windows NT 4.0 has pioneered such as asynchronous I/O, completion ports, and fine-grained kernel locks. These architecture constraints limit the ability of Linux to scale well past two processors.

    All of these said boundries have been since lifted and are really no big deal either..

    Either way, as a UNIX programmer I have to say that I don't see Windows as a poineer of, say, asynchronous I/O, et cetera... :P

    Oh, and that article is all a bunch of FUD anyway. Comparing UNIX to NT is like comparing a Stephen Hawking's brain to a rock. And any operating system that has GUI running in kernel space and can't even run headless is clearly not a server.

  • Re:Well, it did mention ONE good point... (Score:1)

    by Anonymous Coward
    No. Sudo allows 'root' to allow other users access to specific commands and those commands only.

    -b
  • Same thing works nicely with quake and win2k. Or just leave the computer on a few hours with an opengl screensaver. Or use tekram's own scsi driver instead of microsoft's and get your disc all corrupted... Or watch IE5.5 suddenly stop loading new webpages, having to restart it.
    Just my experience win Win2k :(
  • I seem to remember that, actually, but I've gotten too lazy on slashdot to post links lately.

    However, there you have it folks, Linux rules on a real test--a mix of static and dynamic content--even on the evil mindcraft test platform.

    Mad props to IntlHarvester for his MLP skillz; mod him up for bothering to find the link!

    And since I'm also too lazy to reply to your other post: if I want a scalable networked filesystem with ACLs on Linux, I'd use AFS, or just use arla, and get it to work with AFS. If I wanted security, I'd use kerberos on top of that. And since I'm at NCSU, I currently do all that anyhow to integrate with the network.

    The NT boxes on the network do the same, but the interface is lousy and not available to the students, the users have to type their passwords in twice and wait an ungodly amount of time for everything, and the software is more expensive and takes up more resources. Hopefully they'll switch those boxes back over, too, now that Linux is a supported platform here.
    ---
    pb Reply or e-mail; don't vaguely moderate [ncsu.edu].
  • Heck, lets go back to flaming ZD-Net, and telling Jesse Berst "I bet you're getting paid by Microsoft to say that"... :)

    Well, yeah; arla definitely needs a lot of work before that happens, but it makes an okay client now; I understand OpenLDAP is good, but I never had to implement it. However, the distributions have the resources to sink into these projects, easily. And back in the day, Caldera would have just paid for the real stuff, from Netscape or Transarc...
    ---
    pb Reply or e-mail; don't vaguely moderate [ncsu.edu].

  • Tell us (without lying): have you ever seen a W2K bluescreen?

    I have.

    Abit BP6, try using the ATA66 controller with an older BIOS. Bluescreen every time.

  • When we start playing the propaganda and PR game with M$, we're letting them distract us from what we should be doing, making better code.

    Not everyone can program.

    On the other hand, if someone is a gifted writer (and one would have to be, to start a discussion with the PR department of Microsoft), she could spend her time better making some of the documentation that Linux based systems are missing. Though there are some great exceptions, lack of free documentation is still a great problem with Linux. I wonder why Microsoft did not pick that one :-).
    --

  • Oh, indeed. There's nothing wrong with testing that, but the out of the box settings often suck for both of them, and anyone doing real work will have to tune something. For instance, you never get security out of the box, and you often don't get speed; the initial settings are pretty newbie-friendly.

    So, yes, it would be good to use the out-of-the-box settings for some sort of ease-of-use test, but not for a real benchmarking test.
    ---
    pb Reply or e-mail; don't vaguely moderate [ncsu.edu].
  • This was inaccurate a year ago and it's even more so now.

    Truths: in a server environment, speed shouldn't be the #1 factor--stability should. Security: security is only as good as you make it, adn only as good as you set it up. Would you rather set up your Linux box as best you know how, or let the techs at Microsoft do it for you?

    Harder to maintain? Maybe--if you never do anything other than *use* the Windows box, and maintian your Linux box by hand. People who don't want to maintain their Linux box by hand do what the Windows kiddies do--wait for someone to do it for them (come out with update RPMs, debs, etc...that's the subject for another rant) and just grab the updates from the people who maintain their distro.

    Apps? Please. I've seen a number of apps & utilities that one could do on a Linux box using standard command-line utilities. Microsoft was writing this for the folks who are afraid to learn anything.
  • "I don't think"

    Yes, you are correct, you don't.

    The administration functions are available through scripting. You can modify the registry, user identities, permissions, network parameters, etc.
  • I'm serving my ADSL connection to my home via a Win2k server using the routing service.

    It is also pretty common to see NT machines acting as firewalls using commercial firewall products.
  • Hotmail has been running Windows 2000 on their web servers for at least the past month.

  • Looks to me like all of the front end servers are running Win2k. source: netcraft

    Netcraft reported they were migrating about two months ago, and it's conveivable they have all been replaced.

    As far as the back end servers. Sounds like sour grapes. I was responding regarding the point that the front end servers were not running Windows 2000.
  • The way that Unisys provides 99.9 percent uptime is by migrating processes from one processor to another when Windows crashes.

    Indeed, as does Data General. Their 99.9% uptime guarantee is only for their cluster products which basically consist of multiple (usually two, sometimes more?) machines in an HA configuration. If one crashes, the other takes over. Sure, they'll sell you a standalone NT box, but not with the uptime guarantee. For an HA cluster, 99.9% is pretty apalling, which puts it all into perspective somewhat...

  • It's acutally not a joke. I've heard of one case where 2000 was used as a router, and it made sense.

    A training institute running courses on Windows 2000 found it more appropriate to use Windows machines as routers. I suppose when you're teaching MS operating systems, it is better to demonstrate pecuilar uses for Windows than normal uses for Routers.

    I forgot what it can do... IIRC, it's not terribly impressive, VPNs, RIP, NAT, a few other tidbits. All the kinds of things you would imagine a small company would need.

    Oh and now you can administer Windows with Telnet! Security and authentication is provided through IPSec and Kerberos.

    I agree that it is a bit obscene, but technical skills cost money. If you have a Win2k guru on staff, you might as well let them do things the hard way, rather than forcing them to pick up a Cisco manual.

  • What might be fun would be to define a set of pages, loads and requirements. Have two teams run off and build a solution. One using a Microsoft solution, the other using a Linux solution.

    Then measure the solutions based on critera such as remote management, scalability, performance at the prescribed load, performance when overloaded, and of course, cost.

    Other things I might like to see would be graphical web development tools and other stuff people have considered Windows to be particularly good at.

    Linux might get slaughtered, but at least it would have merit.

  • I work in an large environment with retarded monkeys in control. Trust me, the NT and Win2K boxes there stay up despite their best efforts.
  • No - MCSE's and their ilk are simply there to improve your salary and to move to the top of the lists when ill-qualified HR departments are sorting the wheat from the chaff. My MCSE doubled my salary within 12 months of me getting it. Therefore it's a good thing. The RHCE will come to do this in the Linux^WRedhat world as well.

    If you're lucky you might learn something. I taught my NT Workstation course instructor about NT's memory management because he didn't understand it. I also spent most of that week making a mess of other people's boxes (which is trivial because most students don't even change the default administrator password). But the SQL instructor was a very clueful guy. I still occasionally use the MSSQL skills I learnt then when I have to do mysql, Oracle, or Sybase ASE work.

    In the Exchange course I managed to play with the Exchange 5.0 key server which I hadn't used before (or since), and that sort of thing can be helpful, as you don't play with all aspects of a product in real life.

    My MCSE: NT 4.0 stream, Exchange 5.5, SQL 6.5

  • Well - the VERY first time I used an NT-4 box I tried to open a file with Wordpad from a floppy. Turns out the floppy was damaged. Got my self a complete reboot on that one.

    Think about that for a second. A damaged file and an editor took out the OS?

    To be fair, I used another NT box for half a year (mostly running an X server to access unix boxes) and didn't have any problems. But that was all these boxes did.

    As for journaling files sytems, when the journaling file systems from SGI, or IBM are considered, and they are out of beta, that is a moot argument. This is happening REAL fast, so that argument also goes away Real Soon Now.

    You want real world examples of Linux reliability. HMMM - Google, and Deja come immediately to mind. I'd also like an answer as to why Hotmail couldn't be run on NT4? Hmmmm???

    There are also several clustering technologies (and I don't think Beowulf, etc should be considered - they need to be web oriented stuff) that are coming along. TurboLinux has such an offering, and has had such for better than a year. You'll just dismiss it because it's from a "small company." I'm afraid that is a non-argument. Such things exist already.

    Seems after consideration of your entire tome here - you had nothing useful to say ;-)
  • you are incorrect. due to the coarse locking in the kernel, linux 2.2 does not scale well on SMP systems. i believe that the performance drop shows up at >= 4 CPUs.

    in an ideal world, you would have a linear increase in throughput/cpu. however, as you throw more CPUs in a system, the latency in the kernel rises. this happens because part of the kernel locks another part (or itself), making any other CPU which needs that data wait for the lock to be released.

    2.4.x will have much more fine-grained kernel locking, allowing better SMP scalability. not sure how many cpus they are aiming for (8? 16? 32??) - if anyone knows off hand, post it. :)
    --
  • And where was IBM when the "Bug free" Windows 95 turnned out

    By the time Windows 95 shipped in 1995, IBM had made the correct strategic decision put OS/2 into legacy support mode and stop pushing it to new customers.

    Why? Because they spent billions of dollars pushing it in 1987, 1988, 1989, 1990, 1991, 1992, 1993, and 1994, and it never caught on. One last marketing campaign (this time against a better opponent) would not have suddenly made it popular.

    Apple never went on the attack. It's users might have, but the company never did. There used to be a site call the "Evangalist" that mostly bitched about how wishy-washy Apple's marketing was.

    Linux is not like Apple or OS/2. It is not competing directly with Microsoft on the desktop (yet). It's used in a server capacity or by people who want a Unix workstation. And in the markets it's strong in, Linux is actually not the underdog at all.

    (Apologies for the strong language -- I know that OS/2 users are sore about IBM's mismarketing, but I can't stand to see history get rewritten that they didn't market it. OS/2 was probably the most overmarketed product in PC history.)
    --
  • Apple attacked Microsoft while IBM sat around and did nothing...

    That's bullshit. 99% of IBM's advertising for OS/2 touted it's multitasking and robustness (in obvious comparison to Windows), and even marketed it as (famously) "A Better Windows Than Windows".

    Apple had one ad with some old men reading "C Colon Back-slash Windows" from a manual and that was a looong time ago. With a few exceptions, Apple has seen themselves as so much better that it's not worth mentioning the other guy.

    Microsoft defines the market..

    Microsoft defines the market for cheap internet servers? Microsoft defines the market for Unix systems? Check again and stop being so defensive.

    --
  • Come on pb, for someone who reads Slashdot as much as you, how could you miss Linux Beats Win2000 In SpecWeb 2000 [slashdot.org]. Even on a 4 CPU 4 NIC box.

    Mindcraft is over -- Linux won. Get on with it.
    --
  • The page claims, "These architecture constraints [in kernel 2.2] limit the ability of Linux to scale well past two processors." This is ridiculous, since Windows has yet to come anywhere near Linux's massively-parallel abilities. Even with relatively small numbers of processors, Linux performs very well.
  • I wrote this a year ago, but the arguments still hold. Check it out.

    A Response to Microsoft on Linux Myths [dolinux.org].
    ----

  • Sounds like that Win2000 machine was setup for educational purposes only. Not the best example.
  • >Check again and stop being so defensive
    What? You open with "Bullshit" and close with "Don't be defensive?" Cripes...
    Me thinks you get this advice a lot... well nows not the time.. be defesnive... this is a deate.. not a tea party...

    >99% of IBM's advertising for OS/2 touted it's multitasking and robustness (in obvious comparison to Windows)

    What IBM shouldn't advertise the strongest points of OS/2?

    And where was IBM when the "Bug free" Windows 95 turnned out worse than OS/2 Warp (Allready taking a huge publicity beating)...
    Given the chance IBM just backed off and hid.
    At the same time Apple went on the attack.
    (I never saw the "Better Windows than Windows" thing...)

    > Microsoft defines the market for cheap internet servers? Microsoft defines the market for Unix systems? Check again and stop being so defensive.

    As the joke gose "Unix.. is that a Microsoft product?"

    Right now in the busness world "Cheap Internet servers" means Windows NT
    Unix market?
    Your trying to be cute.. Unix means high end servers... thats owned by Sun and Linux dosn't compeate...
  • I say we run a test on a 486 running Windows 95 with 16 meg ram
    vs Linux running on a quad Xeon with 1 gig ram.

    disable the 486 CPU cache...

    Can anyone come up with some tests we can do that Linux dose BLINDINGLY FAST and Windows CRAWLS at?
    I know some of you would say "Anything" but I mean.. something Windows dose especally bad that Linux dose especally well...
    (I don't use Windows byond trying to understand it so I don't know what makes it slow down)
  • >But even Linus has said that LINUX is not as good as the NT/2000 platform from MS in some categories.

    This is a trueism that really isn't bound to Linux or NT...
    Dos is better than Windows 9x in some areas..
    CP/M is better than Windows 2000 in some areas.

    One must rember that to cover one area you need to ADD feature to cover annother you must remove same.
    Linux is getting closer than anyone else to having all the features you'll need and the ability to throw them all away..
    But that is a very very very long ways off..

    In the mean time... I can not imagin what I'd use NT for...
  • I'm familure with both psycology and marketting..

    > The bad side? A tarnished Linux. Why? Because smart and curious people are wary of evangelists of any stripe.

    If this were true then the computer industry is compleatly devoid of thies trates..

    The survival of Apple due to Mac advocates..
    The only reason the Amiga name means anything today is becouse of Amiga advocates.
    The only reason Linux exists today.. Linux advocates.

    The reason OS/2 Warp is dead.. IBM did NOT evangelise..

    Slashdot itself is where it is by advocating....

    The entire computer industry is an industry of evangelists...
    From Sun Microsystems to Commodore...
    This applys to cars as well...
    I've found WV Bug and Rabbit owners to be increadably rabbid.. They remind me of Linux advocates...

    What a sad shame thies advocates have sooo tarnished the WV Bug...

    Yes people get emotional and fanatical..
    Emotional attachment to tools?
    I've known people who'd kill over tools...
    More than that...

    Try saying "Craftsmen sucks" to someone who has a set of Craftsmen tools...
    You'll see seathing anger byond what you'd see in any Linux advocate.
    Part of that is attachment.. part is just plain annoyence..

    Let's say we are both looking at a smart sexy and talented woman..
    Now I say she is ugly and dumb as a rock and can't do anything...
    Again.. annoyence.. You don't know this person.. nither do I..
    We just don't like it when people advocate as true what we ourselfs know to be untrue. (Even if we are wrong).
    Often we justify our anger as just dispising lyers but the reality is often we purely wish to defend our own reality.

    And I have found this holds true in all circles.
    The people who really find Microsoft bashing annoying are people who prefer Microsoft Windows.
    Users not attached to Windows in some way (emotionally, finantually or by friendship) aren't bother by it.
    People bothered by Linux advocacy are those attached to ANY other Os. You need not be attached to Microsoft to be annoyed by Linux advocates. BSD users are probably FAR more annoyed as BSD really is a lot more reliable than Linux.

    Now to add.. Don't confuse annoyence with Linux advocacy with annoyence with people who won't shut up. Everyone is annoyed by an idiot who dosn't know when to drop it and advocates are really good for that.
    On one side I've rightfully had a boot placed to my skull over not dropping the subject..
    On the other.. I've had to be polite far to often to far to many Microsoft advocates who allready know I'm an admitted Linux Zellot.

    But most advocates learn to SHUT THE FSCK UP..
  • > Look at Apple - for years they flogged this whole Apple vs. MS thing, until they realized that it wasn't selling any more computers.

    Apple attacked Microsoft 10 years ago...
    10 years later.. aside from Linux.. they are all thats left.......
    And Mac advocates continue to bash Microsoft to this very day...
    You don't survive by ignoring the company your compeating against...
    Microsoft defined itself by Digital Research in the early 1980s and by Apple in the early 1990s.. now they define themselfs by Linux...
    We can only return the complament...
  • Have you even USED Win2k?

    I tried to install it on my laptop, but it hung part way through. I am still arguing with my hardware vendor about getting an update to allow this install to work.

    When I read that 'Linux Myths' page I had to laugh. I've been able to install just about every operating system including a variety of BSD's and Linuxes on my hardware EXCEPT W2k. Better hardware compatability? Yeah, right - I can't even get it to install.

  • total cost of ownership (for a server solution) includes money spent on administration, and os repairs. Sometimes, this argument does have merit (for example, microsoft would argue that it takes less time to set up an NT box), though for a skilled linux/unix admin, linux should be quicker...

    And if you have to do remote admin, Windows is the biggest pain in the rear end you can imagine.

  • Actually, total cost of ownership (for a server solution) includes money spent on administration, and os repairs. Sometimes, this argument does have merit (for example, microsoft would argue that it takes less time to set up an NT box), though for a skilled linux/unix admin, linux should be quicker...
  • You are absolutely right. Documentation is a big problem. It is even more of a problem with the *BSD's. I've been playing around with NetBSD on both a PC and a Mac Classic II (don't ask). Finding docs on BSD is relatively easy. However finding docs on NetBSD, especially the Mac68k port is not so easy. Hmmm, maybe I should start writing some docs myself. Both for Linux and the other free Unices out there.

    Lee Reynolds
  • OS/2 would still count as "commercial quality", which still refutes the M$ claim...

    As if "commercial quality" means "good" in the first place.
  • Reality: Linux Needs Real World Proof Points Rather than Anecdotal Stories
    >> hmm, that's a cool refutation
    This is just IMHO but if the "Real World Proof Points(tm)" they refer to are the usual overly-focused benchmarks, I'd take a smaller grain of salt with the anecdotes.

    Remember that there is little difference between "Anecdotal Stories" and "Real World Proof Points(tm)" in the first place. They are simply different loaded terms to describe the same entity.
  • It seems like KDE & Gnome focus too much on designing a solution that mimics the Win32 interface. It's kinda like saying "We're emulating the Windows interface because it's the best."

    It also means copying the shortcommings of Win32. Specifically expecting the end user to configure things they shouldn't really have to. Just about ok for a single user machine, but a big pain on a LAN
  • in an ideal world, you would have a linear increase in throughput/cpu. however, as you throw more CPUs in a system

    This "ideal world" assumes infinite bus bandwidth, infinite memory speed and cache managment happening in zero time.
  • my NT server boxen stay up for months on end, just don't get a complete idiot looking after them and they're fine.

    Except that the major selling point of NT is that it's ment to be possible to be admined by a retarded monkey.
  • Ha. I'm an MCSE and still can't figure out how NT works sometimes.

    Is actually understanding how it works the point of an MSCE anyway...
  • The Linux community defends itself because it is attacked. If MS did not attack linux then there would be no need to defend it. All they have to do is to stop lying. Someone once told me its actually is possible to do marketing without lying so maybe MS can try that for a change.

    Posting anonymously because all posts critising MS get moderated down.


    A Dick and a Bush .. You know somebody's gonna get screwed.

  • "Posting anonymously because all posts critising MS get moderated down. "

    I meant to say post anonymously damn that preview!

    A Dick and a Bush .. You know somebody's gonna get screwed.

  • Look at Apple - for years they flogged this whole Apple vs. MS thing, until they realized that it wasn't selling any more computers. Much of Apple's success of late is due to the fact that it no longer tries to go head to head with Microsoft, or define the Mac simply as "better than Windows".
    Cringly recently pointed out that Apple is a boutique computer company. He likens this to other boutique companies such as Porche within the auto market. This is all to point out that investors shouldn't link Apple's fortunes to other industry leaders such as Dell.

    Of course, despite Mr. Cringly's suggestion, this points out that Apple doesn't compete with just Microsoft. Apple is both a software and hardware company. It essentially competes with the entire Intel/PC industry for marketshare. That's pretty stiff competition.

    If anything, THIS is why Apple shouldn't be so focused on Microsoft. And this is where Cringley is right - Apple is its own market. As he points out, you don't see Porche fret over Chevrolet.

  • On another note...
    Look at Apple - for years they flogged this whole Apple vs. MS thing, until they realized that it wasn't selling any more computers.

    ...

    The linux community won't be able to see AOL, Symbian, Palm, or other worthy competitors coming if it mindlessly pursues this David vs. Goliath thing (that nearly killed Apple).

    I agree with one point - there's more than Microsoft out there as Linux competition. Although that's more RedHat's problem than the Linux community. Still, blind advocacy is foolish.

    Having said that... advocacy IS important. And it has helped Apple. How? Lets look at one case...

    Johnson Space Center (NASA) underwent an attempt to standardize their environment. The IT Department suggested various cross-platform data formats. Their director ignored the suggestions and mandated Windows (Windows 95 at that).

    JSC was already a very diverse environment with a strong Apple userbase. Those users were adament supporters of their chosen environment and faught the Windows push.

    Jihad. It even ended up with a congressional investigation. Not because Apple cried fowl - but their users did. Advocacy at work.

    The outcome today? For business automation (read: desktop apps such as email and word proc), JSC is a strong Windows shop. But its still a diverse environment - despite attempts to eradicate all non-Windows desktops. New Apple G4s are being rolled out in increased numbers within JSC. And its all due to advocacy.

  • I've been tracking new NT vs. Linux benchmark results; you can find links to them at http://www.kegel.com/nt-linux-benchmarks.html [kegel.com].

    I've also been tracking the Linux-Kernel mailing list for notes about bottlenecks that probably affected the original benchmarks, and their resolution; see http://www.kegel.com/mindcraft_redux.html [kegel.com].

    Finally, for server programmers, I've been collecting a page of tips on how to write high performance server software for Linux/Unix; see http://www.kegel.com/c10k.html [kegel.com].

  • The biggest problem Linux, BSD, and Apple all face is the fact that "a foolish consistency is the hobgoblin of little minds." For example, my shop now requires everyone to use a standard Win2000 machine (incidentally, the switch was forced before they actually had licenses). I have had arguments because I prefer to do my text editing in Notetab Light instead of Notepad. (I had to prove that formatting problems Notepad was showing could be sorted out by Wordpad as well as Notetab Light.) Even our graphics artist has an idle Win2000 machine sitting under his desk while he does all his work on the Mac Powerbook the company also bought for him.

    The reason for this is an irrational fear my new boss has about anything non-Micros~1. You won't convince him (I believe its because he feels he would be at a disadvantage in a non-MS shop and might lose his position.)

    He's the same way about programming languages. The guy who he replaced was a staunch old time UNIX guy and Linux advocate. Well, some of the computers we inherited from that time are still running Linux, and they have various useful system utilities written in TCL. One of the tasks I was recently assigned was replacing these TCL scripts with Java programs that would do the exact same things. Why? Because my boss knows Java and doesn't want to learn anything about TCL or any other language except for Visual Basic.

    Bill Gates is potrayed as a Borg for a reason you know, Micros~1 doesn't mainly sell operating systems, business applications, games or programming tools. No, what they really sell is conformity, they've figured out how to harness the great forces of social inertia and fear of innovation for their own benefit.

  • You've never run "halflife," then. Recipe for hanging Windows 2000:
    1. boot Win2k (server, in my case)
    2. Start halflife
    3. go into 3d-rendered portion of game
    4. hit the "windows key."
    5. lock up


    ________________________________________
  • then do the immediate apparant thing to do and download newer display drivers that aren't buggy...

    er... like the ones that came on the Windows 2000 CD? Because those are the ones I'm using.

    ________________________________________
  • I notice you use the words "hasn't crashed yet", not "hasn't had any downtime". Not quite the same thing, short of a kernal update, there is no need to ever reboot a Linux-based machine. Not so with Windows products...

    Spyky
  • MS's LinuxMyths [microsoft.com] says: Linux does not provide support for the broad range of hardware in use today; Windows NT 4.0 currently supports over 39,000 systems and devices on the Hardware Compatibility List. Linux does not support important ease-of-use technologies such as Plug and Play, USB, and Power Management.

    (Note that Windows NT 4.0 didn't support Plug and Play either. A curious instance of slamming Linux for missing something that NT was also missing.)

    Ironically, with Windows 2000, Microsoft's offerings have gotten worse in some dimensions; the hardware compatibility list has shrunk roughly in half [microsoft.com].

    The application story for Windows 2000 is similarly weaker since Microsoft has traded off backwards compatibility in an attempt to improve stability. A significant fraction of Windows NT 4.0 applications don't run under Windows 2000 due to new restrictions imposed to improve stability, for example, restricting applications' ability to write freely into the registry (oops, now there was a security/stability hole!).

    --LP

  • Having experts on each side involved in these bake-offs is the most important thing if you're talking about servers that will have a crew to maintain them, but it would also be interesting to see how a moderately technical user would fare with each system, right out of the box. While that seems like it favors MS at first blush, because the GUI is more advanced, if you were to keep track of the costs involved, you might also see the guy with the MS server having to pay more for support, or keep track of licensing issues and crap like that, or just plain not having as much fun in his work.

    Maybe you'd have to do the non-expert bake-off as a 3-6 month test, where two moderately technical users keep a diary of running a small business server. Give both identical Dell hardware, only one gets Windows 2000 and one gets Linux. I imagine the Linux guy would have to learn a little more and do a little more right at the start, but would see that pay off again and again down the road. A big plus on the Linux side might be the fact that the Linux guy just didn't have to be involved with MS. That's the kind of stuff that comes out in a long-term test. How happy are the two guys with their Dell servers, six months down the road? I know there are people who actually like using Windows, but mostly it seems like Windows users hate MS even more than the rest of us, but either they don't have an alternative, or they think that they don't have an alternative.

  • For their on-board LAN solutions on their servers, Dell uses Intel 10/100 (i82559). I believe that Dell uses 3Com for on board LAN on desktops, but I have no experience with them.
  • I don't think you can REALLY administer Windows through telnet

    Actually, you can administer W2k through telnet... I believe that MS is finally taken the proper approach to GUI design for administering objects (whether they be operating system objects or services/daemons)

    The design is basically that they expose a COM or DCOM interface for the object to be administered, and the GUI manipulates that COM/DCOM interface to administer the object or service.
    The obvious advantage to this is that you can perform the exact same administration from the command line as you can from the GUI client through jscript / vbscript / perlscript (and with the .NET platform, you may add C#, etc).
    No more tasks that may "only" be performed through the GUI (a big complaint that I had about NT 4)

    -jerdenn

  • The people who are interested in say 'look at alternatives to windows' are unwilling to do the WORK to benchmark the alternatives. And, few of the people who shout 'alternative' mean 'alternatives', they mean 'their OS of choice'.

    1) Design what/how will be tested.
    2) Create an ISO such that you can place it into the machines on a network so you can pump HTTP queries at the box under test. Transfer the info back to a central logging box.

    By making the process OPEN and able to be done by nothing more than burning a CD, and them booting that CD (Thus not needing a install/reinstall) others will be willing to take a weekend at work and DO benchmarking.

    At the end of day, won't the Linux camp be suprised that not only will Micro$oft beat them in some situations, but BSD also beats them in more situations.
  • However, I'd be interested in seeing another round of benchmarks, especially between Linux 2.4.0 (preferably whenever it's officially released) and Windows 2000.


    On identical hardware, linux 2.4 with Tux web server beats the crap out of Windows 2000 with IIS Click here for result [specbench.org] and don't forget to click here also [specbench.org] for comparisons with different numbers of CPUs/NICs.

    For file system spec test, it isn't pretty. Reiserfs home pages list comparisons against ext2 - they are quite comparable. NTFS hardly wins ANY performance test against any file systems.

  • Please, the results on this page are from machines not even running the same hardware!!!

    How is this a valid test?

    Please check both the links.

  • Could the start of this thread be modded "overrated" once or twice, please? It doesn't have to get pounded all the way down, but it's emotional and tangential to the topic. I came in hoping to read about the issue, not the meta issue.
  • This whole debate is silly.

    If the debate is silly, why do you keep engaging in it?

    If your website is busy enough that the performance issues between IIS and Apache actually matter, then your website gets serious traffic, and you should be way overcompensating on servers in the racks just to catch access spikes and unexpected server downtimes.

    this is just false. Your website may be busy enough that it matters only during those performance spikes, and that's not a rare case at all if you consider basic statistical assumptions (normal distributions, poisson processes, etc.) And if you are going to fill racks, the question still remains as to what to fill them with.

    You may not like the results. Most of the enterprise -ready database products for linux are still nowhere near the performance of even their NT brethren, let alone Solaris.

    Other people shouldn't test things both because they won't like the outcomes and because you already know the results, and the whole debate is silly? Please. Some people are interested in discussing the design of tests and the limitations. You have some things to say about it, but you should do so more constructively, IMHO.

  • You are the first post to actually look at that stale page and say anything about it's contents. Others have cried out that no one needs to look at the page. Some have even pointed out the late 1999 date, as if old FUD was somehow less harmful than new FUD.

    Where did this Ars Fartsica and thes other MicroTurd appologist come from, and why is their prolific garbage ratted so high? It was hard to wade though all the trash to get here. It kind of sucks that all of it is going to be indexed...

  • Actually, I would suggest no tuning done for the non-expert class testing. Just straight out of the box with all defaults. Install any patches the distro and Windows may have and leave it at that. That is probably a more accurate test on the non-expert.

    If you start to include the learning curve of the individuals, you add too many variables to the testing and muddy the results.

    Also, you might want to limit the tuning time for the experts. A single 8 hour period from start of install to completion of task. You can spend all the prep time you want planning how to tune it. Just limit the amount of time on the box. Most places I've been at say they want the server running within a certain period of time. I don't have as long as I would like to tune the box once the hardware has arrived.

  • Comment removed based on user account deletion
  • Didn't motorola come out with a 5 nines linux for embedded systems, hang on....searching...

    oh yeah, it's called HA-Linux (HA==High Availability) [mot.com].

    I chalenge any one to show me 5 nine M$windows, From my personnal anecdotal experience, I would find it too hard to believe.

  • Myth: Linux is more reliable than Windows NT

    Reality: Linux Needs Real World Proof Points Rather than Anecdotal Stories

    >> hmm, that's a cool refutation

    The Linux community likes to talk about Linux as a stable and reliable operating system, yet there are no real world data or metrics and very limited customer evidence to back up these claims.

    >> That would be a good thing to test then. The "uptime" command should come in handy.

    Windows NT 4.0 has been proven in demanding customer environments to be a reliable operating system. Customers such as Barnes and Noble, The Boeing Company, Chicago Stock Exchange, Dell Computer, Nasdaq and many others run mission-critical applications on Windows NT 4.0.

    Linux lacks a commercial quality Journaling File System. This means that in the event of a system failure (such as a power outage) data loss or corruption is possible. In any event, the system must check the integrity of the file system during system restart, a process that will likely consume an extended amount of time, especially on large volumes and may require manual intervention to reconstruct the file system.

    >> JFS, and ReiserFS, should solve that problem really good. And JFS is commercial strength, cause it was in AIX.

    There are no commercially proven clustering technologies to provide High Availability for Linux. The Linux community may point to numerous projects and small companies that are aiming to deliver High Availability functionality. D. H. Brown recently noted that these offerings remain immature and largely unproven in the demanding business world.

    >> Not to sound like a troll, but BEOWULF :p... that's clustering. Or a load balancer...

    There are no OEMs that provide uptime guarantees for Linux, unlike Windows NT where Compaq, Data General, Hewlett-Packard, IBM, and Unisys provide 99.9 percent system-level uptime guarantees for Windows NT-based servers.

    >> Well, that one's sort of true, though there are startups forming. Someone find some examples!
  • Comment removed based on user account deletion

  • That moderator must be drunk too.

  • Dell sells cheap, poorly built machines. If you ever maintained more than 20+ Dell workstations per group, then you know what I mean.

    My department maintains a lot more than that, and I've found them to be a lot better than most of the competition. They use better components, and everything's nice and standard.
    --
  • The claims are too vague to be useful anyway, but going through them anyway...

    Myth: Linux performs better than Windows NT
    Reality: Windows NT 4.0 Outperforms Linux On Common Customer Workloads

    By redefining "common customer workloads" that'll be true. I can also do that to show that my TI-89 is better than Linux. (See, I couldn't get bash to integrate 1/x - it must suck!)

    Myth: Linux is more reliable than Windows NT
    Reality: Linux Needs Real World Proof Points Rather than Anecdotal Stories

    No shit! But then again, so does Windows.

    Myth: Linux is Free
    Reality: Free Operating System Does Not Mean Low Total Cost of Ownership

    I dunno about that one... Windows has so far cost me $200 for Win95 full and Win98 upgrade. Linux cost me $12 - $10 are actually dues to my local LUG, and $2 are a rough estimate of CD-R's burnt.

    Myth: Linux is more secure than Windows NT
    Reality: Linux Security Model Is Weak

    This is actually quite true, but it's changing (dunno about in 2.4). The security model has one all-powerful user, and no permissions. I *know* from looking at 2.2.10 code that people were starting to implement user permissions/permissions for various kernel tasks (ie, a process can be allowed to update the system clock but be denied access to bind to ports lower than 1024). The file system has permissions separate from the kernel, sorta, and... ugh.

    Myth: Linux can replace Windows on the desktop
    Reality: Linux Makes No Sense at the Desktop

    I can redefine "desktop" to make this true. Actually, in this reguard, I'd say that Win2000 is the best OS, in that it gets a nice balance between "desktop" tasks and "server" tasks. Then again, I don't run Win2000, I dual boot between Win98 and RedHat 7.0. Desktop under Linux works for me. But then again, I do most serious office-like tasks under Windows. This is sorta because lpr for whatever reason only works if I'm root and also because last time I tried to print under Linux, only the top row of the 600x600dpi matrix actually printed.

    The Linux operating system is not suitable for mainstream usage by business or home users. Today with Windows NT 4.0, customers can be confident in delivering applications that are scalable, secure, and reliable--yet cost effective to deploy and manage. Linux clearly has a long way to go to be competitive with Windows NT 4.0. With the release of the Windows 2000 operating system, Microsoft extends the technical superiority of the platform even further ensuring that customers can deliver the next generation applications to solve their business challenges.

    For the time being, I'd have to agree. Linux has come a long, long way - but for most things, Windows is currently better. (My current pet Linux peeve is the brain-dead scheduler... this mostly has to do with the fact that it isn't pre-emptive. Of course, I have a different beef with the WinNT scheduler about starving processes... Dunno if Win2000 has a different scheduler, but I'll bet a single CPU intensive task will make the thing look like it's crashed. I can explain my take on both if anyone wants more facts.)

  • Since it will be lost in the noise, but as soon as I read the title of this post I knew that the forum would be filled with "No, Windoze sux cause we all know that LINUX is better." Windows 9x sucks...yes, this is true. But even Linus has said that LINUX is not as good as the NT/2000 platform from MS in some categories. Perhaps when 2.4 is out this won't be true anymore, but it does little good to sit here and accuse any site that says LINUX isn't the end-all/be-all solution of spreading lies.

  • Static web content. This is almost a non-issue for dynamic eCommerce sites that generate most pages from database content

    This whole debate is silly.

    If your website is busy enough that the performance issues between IIS and Apache actually matter, then your website gets serious traffic, and you should be way overcompensating on servers in the racks just to catch access spikes and unexpected server downtimes.

    Servers need to be tested for performance talking to Oracle/DB2/SQL Server databases

    You may not like the results. Most of the enterprise -ready database products for linux are still nowhere near the performance of even their NT brethren, let alone Solaris.

  • Apple attacked Microsoft

    Yes, it was an amazingly dreadful display of pissing into the wind that was the hallmark of the Spindler/Amellio administrations, and nearly put Apple out of business.

    Look at any Apple marketing now and you will not find any direct comparisons to Microsoft products - this has little to do with the $150 million investment MS made in Apple and more to do with the common sense that it was just lousy marketing.

  • Again, whatever Dell said about Linux in November, 1999 has to be regarded as stale. Dell's page (link kindly provided by the Microsoft Corporation) says this:

    Borrowed from: http://www.dell.com/us/en/dhs/topics/linux_linuxho me.htm

    Thinking about Linux and how it will play in your high-tech environment?

    You're not alone. Lots of people are looking into Linux as a low cost, stable Operating System alternative. Linux is becoming an important part of the Internet Infrastructure with an impressive 33% of web hosting servers running Linux - and that number is growing. Linux is also widely deployed on file and print, e-mail and application servers and on workstations for application development.

    Dell and Red Hat have teamed up to deliver the most comprehensive menu of Linux based products, services and solution stacks available.

    Announcing the Dell and Red Hat One Source Alliance

    Dell and Red Hat have combined forces to form the One Source Alliance, a unique relationship that unites Red Hat's leadership in the Open Source market with Dell's commitment to meeting the demand for Linux from the edge of the Internet to the heart of the Enterprise. Our goal is to speed the commercial acceptance and adoption of Linux and to provide one source for Linux solutions.

    The One Source Alliance includes collaborative development of products, co-development and delivery of global service offerings, and solutions that are enterprise-ready, Red Hat certified and offer the best customer experience. Together, we are also giving back to the community by developing tools and test suites that have been made open source and are readily available.

    Through the Alliance, Dell and Red Hat intend to work together to promote the adoption of Linux, offer the services and solutions the community is looking for and drive future technology innovations to support the infrastructure computing needs of the Internet economy.

  • Challenge accepted. Stratus [stratus.com] provides 5-nines availability for Windows 2000 Datacenter. (BTW -- Windows 2000 hasn't crashed on me in 8 months of use).

    As was previously stated, these guarantees are more PR than anything but you brought it up....

  • It needs an update, but for Windows 2000. How does the NTFS5 system perform? What about Active Directory? The new command-line tools?
  • Linux security is all-or-nothing. Administrators cannot delegate administrative privileges: a user who needs any administrative capability must be made a full administrator, which compromises best security practices. In contrast, Windows NT allows an administrator to delegate privileges at an exceptionally fine-grained level.

    I never particularly liked the idea of one super-uber user. It doesn't make sense. This has to be the biggest annoyance of Linux.
  • An NT router?

    Is that a joke? Please dear god tell me that's a joke!

  • Actually, I'm running Windoze, BeOS, and Linux and I have to say that there's no real comparison between Windoze and Unix/Linux/everythingelse.
    You can kinda think of Windoze as an OS for the general public with little or no experience on computers or someone who just plain doesn't want to know and expects everything to be cut and dry.
    Linux, on the other hand, is not as "layman- freindly" and is more for those who enjoy having control over their computer and software; creating and manipulating data.
    Guess it just kinda depends on what mood you happen to be in, eh?
    --

    Vote Homer Simpson for President!

  • Folks, the propaganda on the Linux Myth page is not only inaccurate and blurred, but also has some VERY FUNNY aspects: Did somebody check the mentioned companies on Netcraft? I just did, and I found out, that the both Chicago Stock Exchange and Boeing DON'T use Windows NT as the platform for their main website! Why do they use Netscape or Apache servers on Unix derivates, when Windows NT is such a stable, fast, versatile, affordable, simply just KEWL platform? Apparently, M$ has to hide itself and it's products behind the propaganda in the Linux Myths paper. Come on, M$, can't you really do better? ix
  • Obviously NT/Linux hasn't gone through the scientific/academic environment speech yet.
    My personal observations from the Physics/Astronomy department at the University of Pittsburgh:
    • Windows 95/98/NT/2000 usage on the whole is dropping like a brick. Every new PC that comes in to the department either dual-boots between Windows [98|NT|2000] and RedHat or runs RedHat exclusively.
      Reason?
    • Professors and grad students here (mostly) love Mathematica and Maple, and find that Linux & Solaris allow them to run much more smoothly than on Windows *. You want stability tests and benchmarks? Use them two packages -- you'll see some pretty convincing results.
    • Professors and grad students here run a ton of Fortran code. g77 comes with Linux. Find a Fortran {77|90} compiler/interpreter/whatever for NT that's stable *and* find a prof/grad that's willing to use it.
    • Professors and grad students around here are notoriously cheap and stubborn. They've grown up on Unix (sadly, some also did VMS, and now have horrid psychological problems because of this) and Linux provides a very familiar interface. You'd believe the number of people here who bring up a DOS box in Windows and type 'ls'.
    • AFAIK, there is no good ssh1/2 & scp client/server package for 95/98/NT/2000.
    Your thoughts?
  • "Since these people can't understand that a computer is simply a tool and nothing more, their opinions should be taken lightly at best."

    "Just a tool", huh? Science is "just a tool" as well--it simply a series of steps and conditions that allows a person to gain accurate information about the world. Tools are very very important. Having access to tools often means the difference between life and death. It certainly means the difference between cave (or tree) people and medical care, shelter, education, etc.

    Computers are becoming the most important physical tool we have (science is largely a mental tool). The store and process our information--something that used to be done slowly and with great error. The are instant communication devices. They are (I hate to use the word) "empowerment vehicles".

    Given the importance of computers, shouldn't we be a little concerned who is running them for us?
    --
    An abstained vote is a vote for Bush and Gore.

  • Alright now. Who here has driven a Chevy? Come on, let's see a show of hands. Who's driven a Pontiac? Who's driven both?

    Have any of you noticed that Chevys keep running damn near forever? I mean you can break nearly every single part in the car and it keeps running! Pontiacs on the other hand - you can zip up and down the road acting like your a Nascar driver or something, getting those instant kicks you think is fun. But you know what? 70,000 miles and those suckers are history. And it isn't a matter of a few repairs here or a little tweaking there, the damn thing's burnt itself apart.

    Same thing with Windows vs. Linux. I can take Linux and run it on an old 386, break half the software on the system and it keeps running. Windows? Pfff. Need a new box every two years. You can't even run the same Windows version as you used to because none of the software supports it any longer!

    This is where Unix truely shows Windows what it's worth. There is actual value in those SparcStations from seven years ago. It's actully quite amazing. Most of those SparcStations will STILL out perform windows at serving up content. Huh. Go figure. ;-)

    One more thing. I am being somewhat inaccurate in my description of Linux as the Chevy of the software industry. That honor befalls FreeBSD or Solaris. Linux is more like Fords. Close, but not quite. (For example, my data STILL isn't safe under ext2) Don't worry though, they'll get it figured out eventually. :-)

  • I believe that there is a solution with an uptime guarantee in linux. Namely the s/390 from ibm which you can get configured for linux. I am fairly confident that that version is covered by the same uptime gurantees as other operating systems. The cost of a midgrade s/390 might be comparable to running a server farm in nt 4 or win2k given 50 midgrade servers might run you $20,000 each for about $1,000,000 total cost which is conservative considering the cost of buying 50 site licences for back office and windows itself (each run $5,000) leaving you with $10,000 to buy all the hardware and any other software you might be running. P.S. I do not work for ibm

  • Re:Linux Myths vs. Netcraft Reality (Score:3)

    by sql*kitten ( 1359 ) on Saturday October 07, 2000 @01:49AM (#724622)
    both Chicago Stock Exchange and Boeing DON'T use Windows NT as the platform for their main website

    That's right - the stock exchange run their trading systems on NT. There's more to the computing world than web sites, y'know.

  • Re:Well, it did mention ONE good point... (Score:3)

    by Mihg ( 2381 ) on Friday October 06, 2000 @09:33PM (#724623)

    Work is under way to fix this.

    Everything that uid 0 has traditionally been allowed to do has been factored into different capabilities (CAP_SYS_TIME allows you to set the time, CAP_DAC_READ_SEARCH allows you to read any file or see the contents of any directory, etc.).

    Each process has three sets of capabilities: effective, permitted and inheritable.

    • The permitted set is the maximum capability set that the process can have. The process can only turn off capabilities in this set. A process can limit itself to the capabilities that it specifically needs.
    • The effective set is used when security checks are performed by the kernel. The process can turn off any capability in this set, or turn on any capability in this set as long as that capability is also on in the permitted set. The effective set allows a process to temporarily enable capabilites to perform some task, and then turn it off later when it is no longer needed.
    • The inheritable set contains the capabilities that are given to child processes when the process forks, allowing a process to restrict what its children can do.

    Right now, the kernel automatically gives uid 0 process all capabilities and removes them when a process changes its uid. However, in the future it will be possible to associate capabilities with executables at the filesystem level (similar to the setuid and setgid bits now) and (along with access control lists) restrict the execution of the process to specific users. This allows the construction of systems where uid 0 isn't special and each user is specifically limited to certain privileged operations.

    The goal is to eventually add Auditing, Mandatory Access Control and Information Labeling to Linux, allowing for a possible B2 (or whatever Common Criteria calls it...) certification for Linux. (Which is something Microsoft will never have.)


    ---
    The Hotmail addres is my decoy account. I read it approximately once per year.

  • Re:Linux Myths (Score:3)

    by Angst Badger ( 8636 ) on Friday October 06, 2000 @11:18PM (#724624)
    tune it all as much as possible

    That's all well and good, but I think it'd be interesting to see how the out-of-the-box setups compare, especially because Microsoft likes to point out how easy their systems are to administer for non-experts versus Linux (while simultaneously raking in the bucks for MCSE training and certification).

    --

  • WHO CARES??? (Score:3)

    by leereyno ( 32197 ) on Saturday October 07, 2000 @01:17AM (#724625) Homepage Journal
    Have we reached the point where we put a priority on responding to fud and propaganda?

    What on earth do we have to prove to Mickeysoft? That Linux is better (or at least competitive) as a server? Anyone with a lick of sense already know this. Does the fact that Mickesoft says otherwise matter? Anyone who would accept M$'s viewpoint on one of its competitors without a huge grain of salt is a clueless moron. Do we have any reason to care about the opinions of the stupid?

    The only way to respond to things like this is with code. Use M$'s pokes and jabs to evaluate Linux for possible weaknesses, and then go fix them. Concentrate on making Linux the best operating system we can. When we start playing the propaganda and PR game with M$, we're letting them distract us from what we should be doing, making better code.

    Lee Reynolds

  • Who cares? Quit defining yourselves via Microsoft (Score:3)

    by Ars-Fartsica ( 166957 ) on Friday October 06, 2000 @08:36PM (#724626)
    This is getting old.

    How long is the /. community going to define itself through Microsoft?

    Move on folks - there is a whole huge world of computing out there and as dominant as Microsoft is (or more accurately, was), the whole pie is going to grow by such a substantial amount in coming years that any notion of "us v. them" is illusory - the IT market, and hence the market for linux products, is expanding, mutating, and growing. TV, cellphones, home automation, handhelds, servers, etc.

    Added to which, the whole linux v. Microsoft thing just gets really tired.

    Its probably AOL you should be more concerned with at this point.

  • Re:Static web serving doesn't count for much.... (Score:3)

    by Ars-Fartsica ( 166957 ) on Saturday October 07, 2000 @07:21AM (#724627)
    If the debate is silly, why do you keep engaging in it?

    Because I have multiple years of experience working at a web site that get well over one hundred million hits every day, which gives me some insight into the issue others may not have.

    Your website may be busy enough that it matters only during those performance spikes, and that's not a rare case at all if you consider basic statistical assumptions (normal distributions, poisson processes, etc.)

    Drop it, you're applying statistical phrases to look smart, but they have almost no relevance to website performance spikes. Every site in the top 500 sites has had significant increases in page views in the last three years - these aren't spikes - you can be rest assured that for almost any popular mainstream site, there is a good chance its pageviews are doubling annually, and your server farm should be growing ahead of this. Hardware is almost always the limiting factor in growth. I have never heard anyone tell me that their site performance improved substantially due to webserver software (by going from Netscape to Apache, etc.), and I'm not talking about Slashdot, I'm talking about some of the busiest sites on the web.

    Other people shouldn't test things both because they won't like the outcomes and because you already know the results, and the whole debate is silly?

    No, if you actually go back and read the post you'll see that the poster was fishing for hints that enterprise DBs on linux are superior to NT. My response wasn't to tell him he shouldn't test, only that I already know the results (from having all of the db and platform vendors give their best shot to win a contract through a multiweek trial).

  • Re:Who cares? The business world, for one! (Score:3)

    by Ars-Fartsica ( 166957 ) on Friday October 06, 2000 @09:17PM (#724628)
    Non-geeks don't know anything other than what the numbers and PR tells them.

    No, you're missing the point here - most of the reasons not to define yourself in terms of one competitor have everything to do with marketing, and almost nothing to do with technical issues.

    Look at Apple - for years they flogged this whole Apple vs. MS thing, until they realized that it wasn't selling any more computers. Much of Apple's success of late is due to the fact that it no longer tries to go head to head with Microsoft, or define the Mac simply as "better than Windows".

    The linux community won't be able to see AOL, Symbian, Palm, or other worthy competitors coming if it mindlessly pursues this David vs. Goliath thing (that nearly killed Apple).

  • Re:Who cares? Quit defining yourselves via Microso (Score:3)

    by VividU ( 175339 ) on Friday October 06, 2000 @09:01PM (#724629)
    Don't hold your breath.

    For reasons probably better explained by phsycologists and sociologists, some people tend to develop emotional relationships to their computer/OS/Software/Newsgroup/ect..

    Remember the Apple vs. MS thing? How about when AOL first opened up their service to the Net?

    Since these people can't understand that a computer is simply a tool and nothing more, their opinions should be taken lightly at best.

    Now it's Linux vs. MS.

    The bright side? A backlash is coming and its coming fast and the sooner the better: People who define themselves by their OS of choice ultimatly end up hurting the cause their championing. They need to go.

    The bad side? A tarnished Linux. Why? Because smart and curious people are wary of evangelists of any stripe.

  • This is a stale Microsoft Page (Score:3)

    by DavidBrown ( 177261 ) on Friday October 06, 2000 @08:48PM (#724630) Journal
    At the bottom of the page, Mr. Gates et. al. kindly point out that the page was last updated on November 1, 1999. Also, it's a comparison between Windows NT Server 4.0, not Windows 2000 Professional (or whatever they call it these days).

    What does this mean? Nothing really, not anymore. It's stale data.

    Maybe it's time for both Microsoft and the Linux community to update the page. As for Dell, it's not that hard to believe they would have said what they said about Linux - last year, before the recent Linux boom and the Microsoft verdict.

  • Linux Myths (Score:4)

    by pb ( 1020 ) on Friday October 06, 2000 @09:08PM (#724631)
    That page wasn't accurate *then*; I'm sure you can look up several refutations of it.

    However, I'd be interested in seeing another round of benchmarks, especially between Linux 2.4.0 (preferably whenever it's officially released) and Windows 2000. Also, a comparison between the many actual journaling filesystems that run under Linux and the one (not really) journaling filesystem that runs on NT/2000 would be interesting; heck, include BeOS in that one, too! :)

    Ground rules: use identical hardware, use stable, recent software, tune it all as much as possible, and test with multiple hardware configurations, (i.e. not just 4 Processors and 4 NIC cards; that's not terribly realistic in the first place) to figure out what is the best overall solution.

    Also, when determining web server performance, make sure there are some tasks that are CPU-bound or IO-bound as well as simply network-bound; you want to test everything.

    That all having been said, I'm pretty sure that a lot of the Mindcraft-specific differences have been fixed in 2.4, and therefore Linux should perform significantly better on those sorts of tests than it did in the past, as well. I'd rather see some more well-rounded tests constructed to go along with that, though.
    ---
    pb Reply or e-mail; don't vaguely moderate [ncsu.edu].

  • Good god (Score:4)

    by GoRK ( 10018 ) on Friday October 06, 2000 @10:00PM (#724632) Homepage Journal
    When are people going to grow up and realize that all architectures deliver or can be made to deliver basically the same sort of general performance and that any OS is going to shine in some particular area at some particular point in time?

    IT managers really get my goat.. Lessay the entire development staff has a UNIX background and wants to develop a web application on PHP and Oracle. IT decides that NT is "better" than linux because ASP can serve pages a little faster than Apache can dish out PHP3 so then what do they have? Ground zero. Development staff that can't work with the tools. Same thing on the flip side... Try to make flashheads/Adobe GoLive! style designers work with Linux webservers. Give it up. Buy NT.

    NT4 and Linux 2.2 kernels offered basically the same thing packaged two different ways. Windows 2000 rocks. Linux 2.4 rocks. Why is it a fight? I like 2000 on my desktop; Linux on my servers.

    Why doesnt someone start a Linux Marketing Fund to buy advertising and pay PR people to dish out the same shit for Linux that everyone else is dishing out for their junk? Why does Linux always have to remain on the devensive? It could have the hell marketed out of it if someone wants to fund it.

    Where are your priorities anwyay?

    ~GoRK

  • Re:Who cares? The business world, for one! (Score:4)

    by d.valued ( 150022 ) on Friday October 06, 2000 @09:11PM (#724633) Journal
    Non-geeks don't know anything other than what the numbers and PR tells them.

    I talk to people who wonder about the "LINUXGRUVEN" sticker on my laptop, and I oft come up with "Windows is easy to learn and hard to use, while Linux is hard to learn and easy to use."

    User-friendly, ergonomic, simple interfaces are for Gee Q. Public. There is an overwhelming majority that use computers for e-mail and IM and basic web work and word-proc-ing. They want a simple interface and a simple system. At least in the first part, MS did it.

    But.. if you offer to teach people on how to use Linux, we can make an inroads. I've offered my cousin, a 13 yo who likes computers and hates AOL (praise Jeebus!), to teach him how to use Linux and let him use some of my sacred pile of O'Reilly books. I offer classmates that are interested the same option, to liberate themselves from Microsoft tyranny (and save big bucks in software).

    No takers, but the offer is open.

    The way to increase Linux use by GQP is to make it less imposing. Teach it to your friends.

    Yes, this approach may sound familiar. It's how most religions got started, by a faithful few spreading the word to the World At Large.

    YES. I am a collar-wearing linux evangelist. I intend to join the GeekCorps.

    Amen. Thus ends the lesson.

  • Static web serving doesn't count for much.... (Score:5)

    by NeoMage ( 29426 ) on Friday October 06, 2000 @08:33PM (#724634)
    I see a few problems with the suggested tests.

    Static web content. This is almost a non-issue for dynamic eCommerce sites that generate most pages from database content and/or personalised content. In the coming age of Internet transactions there will be little requirement for purely static serving at large volume. Servers need to be tested for performance talking to Oracle/DB2/SQL Server databases, and also server to server XML. These are soon to be the more prominent roles of the web server in the business world rather than just serving up static pages.

    And routing performance against NT? Why bother? How many people do you know that buy NT boxes for routers? If you're going to spend the money you might as well buy a hardware solution. If you don't want to spend the money, then everyone knows Linux already makes a decent router regardless of the 2.4 kernel.

    When it comes to a purchasing decision, it will end up coming down to more than just this sort of testing.

  • Strangely enough, there are some good points here (Score:5)

    by Ars-Fartsica ( 166957 ) on Friday October 06, 2000 @08:57PM (#724635)
    Windows NT 4.0 has been proven in demanding customer environments to be a reliable operating system.

    That might sound like marketing-speak, but for better or worse, its true. NT is there in the server farms of America, and it is staying up and performing in spite of itself. Get used to seeing Win2k more and more in the server farms you visit in coming years.

    There are no OEMs that provide uptime guarantees for Linux

    Once again, the grain of truth here is that none of the linux distro vendors can really be taken seriously at this point, and most have thrived purely on the goodwill of the community. I can't think of one commercial distro that is truly enterprise worthy outside of Debian (which I consider noncommercial).

    After the RH 7 debacle, do you really expect them to offer uptime guarantees? Sometimes I wish someone like IBM would issue a distro, just so there would be a real player with a serious service commitment behind the product.

    Linux is getting there, and for most of its users its presents an incredible value. You have to look past these customers at some point and consider how you are going to satisfy customers who do not have budget constraints, but do have significant demands of the software and support they pay for.

Slashdot Top Deals

Thus spake the master programmer: "After three days without programming, life becomes meaningless." -- Geoffrey James, "The Tao of Programming"

Close