Slashdot Mirror
Linux Advocacy Hurts
sundae writes "For those who are into Linux advocacy,
check
this little piece from WinInfo before you
pointlessly brand any critizism as FUD, then attack the
author and everything else related. This article criticize the
way many pro-Linux sites, such as /., only whine about how
the Mindcraft study was funded by Microsoft and ignored the
fact that the study actually *did* happen. "
My opinion is that the Mindcraft study is completely invalidated
by using an optimized Windows box and a vanilla Linux box. Its
that simple. I'm kinda bummed that this article fails to
mention this, instead implying that the Mindcraft test
was reasonable. But don't miss the point- we're our own
worst enemy. I got CC'd on a lot of criticism of
this report, some of you guys did an excellent job
of fairly ripping the report to shreads. Others... well...
didn't.
They went OUT of their way to make Linux/Samba/Apache perform badly.
This article is full of FUD and logic flaws. I fail to see how you can think this is a "great article.
How is putting Apache in inetd more secure? How is using widelinks more secure? How is using a kernel with documented flaws more secure? How is running nfsd on machine not needed NFS more secure?
Instead of shooting your foot, why don't you be more productive and go read the report for yourself? It's all there in black and white.
First, I don't know if more people do use Linux servers. Second, the premise that because something is used more widely it must be the best is a fallacy. If it were shown that NT was used more than Linux would you blindly assume that NT was better?
Only smart people use OpenBSD -> Most Linux users are stupid. More people use Linux servers = Linux is better, right? Stupid argument. What OS you use SHOULD depend on what needs you got, not everyone wants or needs Linux.
I did not read the article by MindCraft... Was
there anything said about the differences in
speed after long periods of uptime? I have found
that linux is fast...real fast all the time. I
have noticed that NT is not slow...at first, but
open Netscape and close a couple of times...
while watching the memory get eaten up...
The article says that any anti-Linux news/articles/postings are listed on Slashdot.org AND is asked, after description, something like "Guys, critiques?", "Anybody has arguments?", etc.
Note that I am NOT saying anything technical, I just don't think that many pro-Linux users are objective.
Guys, critiques?
This was not a 4-way system. Who's the dope?
Doesn't putting stupid values in Apache's well-documented conf files count as going out of their way?
I can't get in! Maybe they're running one of those slow Linux boxes with apache? :)
Here they say that they couldnt find but bits and pieces on the apache server, even though im overwhelmed with information. What i wonder is, why do they try to do a benchmark themselves, if they do not know how to use apache or any part of linux for that matter. I would have believed the test if it was tested with the help of someone who actually knows what they are doing. Just another thing to point out, thats all.
Another fact is that NT DOES beat up on Linux in 4-way systems.
And where exactly do you get your "facts" from? Surely not the Mindcraft report?
Nah, thats not really correct way of doing a comparision either.
:-)
:-) It is better choice for me, on what -I- am using it for, but I believe that there are people and situations where something else can be a better choice, it all depends on the task, the applications, the hardware, and the people setting it all up and operating it...
Just because lots of people are doing something doesnt mean they have to be right about it. There are so many people in this world, who just follow others and repeat what they hear from others without thinking themselves. The most bought and used products are usually the ones who had the largest advertising budgets, based on that the products are fairly OK from the beginning, that is.
Now, I dont think this applys to GNU/Linux of course straight on.
GNU/Linux is excellent, whether or not lots of people are using it. I assume you are one of the "newer" linux users, who started using GNU/Linux after it became popular to the big mass, otherwise you would probably be thinking more like I do on this. I have been using Linux for years now, and I dont think is was that much worse a few years ago, in comparision to other OS's , just because less ppl knew about it back then
So,
More users != it is the better product
GNU/Linux just happens to be the better one anyway, based upon its techical capabilities. Although, that depends what you want to use it for
That is a fairly dangerous path to go down, it would be countered by "More people use windows than linux, therefore by your own argument windows is better" ...
Do not be fooled. They knew exactly *what* they were doing and *why*. These people know Linux even better than you think.
So what you're saying is you don't know how to administer NT?
It's a poor workman who blames his tools.
Bill would like y'all to forget about the one benchmark that matters to him... cost.
Please stop sending me email.
/. reader, not the /. effect.
Well if you want people to stop sending you mail, then make is clear as day. I personally think he went a little overboard on the article if he wanted people to stop sending him mail.
If anything he added more fuel to the fire. If he wanted to do the smart thing, he would have put a link to the ZDNET article which showed that the tests were biased and are invalid. Instead he insists on shooting himself in the foot.
I'll pass on emailing him because I respect his request, but I represent one
simon
If the testing methodolgy is sound, what does the source of the money matter?
You must attack validity of the testing methodology.
-----------------------------------------------
Domain: sorehands.com User:bill
http://www.sorehands.com
-------------------------------------------------
Fired by Microsystems Software (now owned by the Learning Company) when he went to the Hospital for RSI treatment.
Cya @ kindargartten, Diabolikal.
I never owned an Amiga myself, but I was in the thick of all that when it was happening (Some of you youngsters might not remember.)
Commidore's death was caused by a variety of factors, including:
1) Inability to market. The industry saw the Amiga as a game machine only. If Commidore had made a point of showing it capable of some very nice business processing, we'd probably all be using them now instead of PC's.
2) Proprietary hardware. This has been the death of many a computer company. Proprietary hardware == more expensive and harder to get hardware. This almost was the death of Apple, too (Still could be, eventually.) Another example of this is IBM's Microchannel archetecture which was plug and play years before anyone even coined the term.
3) Lack of developer support due to 1) and 2).
The C= Amiga was years ahead of its time. A friend of mine had one in college and the graphics and speed of the system were not seen on the PC side of things until years later. It wouldn't have taken much for things to turn out radically differently.
Every OS has a lunatic fringe that invariably embarasses the rest of us in the advocacy department. I can not count the number of times I had to go in to do damage control on the OS/2 newgroups due to some idiot shooting his mouth off.
I guess you don't see them as often in relation to Windows due to the fact that the OS doesn't need a lot of advocacy. That may change as Linux and other OSes encroach more on the Windows territory. I'll be amused when we start seeing these yahoos pushing Windows. I suspect also that the age demographic on die-hard Windows users is significantly higher, and they're much better at making statistics work (*cough*lie*cough*) for them (Age and experience will always defeat youth and vigor?)
If they know alot about linux, why are they digging the web for information on the apache web server? Of course then again, theres a big disclaimer on the bottom saying theyre not liable for false tests. Personally, i help my high school get a network set up, and when i went to linux from nt, the crashes when everyone was on a computer disappeared. But thats just from my experience.
Two identical machines. 1 Windows team, 1 Linux Team, 2 Hour setup.
I'm saying don't underestimate your foe. They knew they were killing performance by the changes they made. Every move was calculated.
You may think Microsoft is stupid, but in fact they employ quite a few certifiable geniuses.
In my opinion, one should not argue too technically when talking about people why to use free software instead of M$. Windows NT is a very good OS, as is Be and Mac OS, too. I think the NT platform is technically much better than the free software stuff that's around. But that's not the point!
The main disadvantage is that it is not free! You always have to pay for it. You have to pay to somebody who does not have scruple to spy you. You can never really be sure that Windows does not send some contents of your harddisk to M$, that it will run even if some software is installed that Bill doesn't like, or that you can open more than 10 outgoing TCP connections.
I would benchmark NT vs. Linux, I would certainly tune much more on the Linux side because its much more fun to work with something that's open and does not try to keep secrets. Well-paid M$ employees certainly have another motivation to work with their nasty protocols.
To dismantle this benchmark, I would go further into
1) Does somebody else besindes M$ have the knowledge to archieve this performance on their broken protocols? Is it desireable to rely so much on an unfair company?
2) How much can you trust companies like Mindcraft or Truste at all? Did anybody verify their results so far?
And yet, nobody expects for a spammer to be the "normal" emailer, or when someone goes to a movie for him to act the same as when he goes to a corporate meeting. Neither should the press, or anyone else, expect that the average slashdotter is a loud-mouthed idiot, or that when someone posts to slashdot, that that is the only thing that he thinks. A person can come accross as trashing microsoft, when in reality, he just has a problem with one or two programs. Then, when people read that, they generalize it into "All slashdotters are microsoft trashers."
GNU/Linux is in a very sad state. Little hardware support. No common API to use graphics/sound/etc. About all you can do is use console text and networking. There won't be an agreed apon desktop to use since development is split between 2+ desktops. Now many people will say this is pure FUD. Maybe you people should try creating something useful with Linux (and use Linux for a little longer than 5 months). Nowadays to use any program you have to download libraries from hell. And the number of libraries is increasing rapidly. The media is helping a huge deal by saying Linux is actually good. But, Linux is not all that.
"Quit bitching and code", I hear you say. Well you try to create a library that everyone accepts. You will get people saying "Why _yet another_ xxxx??". People will say "use library xxx", but that library is nothing but crap. 80% of Linux users will not have the library anyways. Every program I have come across has reinvented the wheel trying to use X/sound/etc. and still be portable. Soon we will have programs for both KDE and GNOME. That will be fun.. seeing many programs with C and C++ mixed together. Wrappers for wrappers even have wrappers today. Xlib is a wrapper for the X protocol, glib is a wrapper for C, GDK is a wrapper for X which sits on glib, GTK is a wrapper/system for GDK which sits on glib, GNOME is a wrapper for GTK (and about 1,023,442 other libraries). Can you say bloat? This is _reality_, not some MS funded FUD. I'm a Linux user and coder. Been that way for years.
Sure, Linux can kill NT anyday in networking.. but maybe thats because Linux is sitting idle more than NT. I'm sure once you get X, KDE/GNOME (after they are "complete"), a window manager, and Netscape and a few other misc. programs running your system will be fairly unusable and network performance will decline.
BTW: Was this test of Linux vs. NT done with Linux running X and a desktop system? I don't see how its remotely fair if Linux was being ran in just console while NT was ran in a GUI (even highly optimized).
I think that you should stick to CS. You obviously a)have never read the book, b)know nothing about industrial psychology, c)know nothing about individual differences, and d)know nothing about statistics. I realize that your political views let you feel as if you know everything, but perhaps you should let us talk about Linux, as opposed to IQ?
>BTW: Was this test of Linux vs. NT done with
>Linux running X and a desktop system? I don't
>see how its remotely fair if Linux was being
>ran in just console while NT was ran in a GUI
>(even highly optimized).
The GUI on NT would just be sitting there doing
nothing, it would probably be paged out.
Besides, the fact that you cannot deactivate NT's GUI is a disability, NOT an excuse. Running X and KDE on the linux box would be as fair as running an X server, along with KDE on the NT box as well.
I think you should stick to giving Myers-Briggs tests. (this is a personality inventory that is very widely used in "industrial" settings but has been academically torn to shreds for some time.)
IQ testing has wide commercial use, but that doesn't mean it's valid. IQ test boosters haven't even come up with an adequate definition for intelligence yet other than "that which the IQ test measures."
Objections to the IQ test on grounds of environmental bias are still being evaluated. Many are quite serious. Don't write them off out of hand.
(for example, japanese performance in america in academics is higher than the white average - before you decide that this means a genetic advantage, look at some other statistics instead of just staring at IQ tests. you'll find that a disproportionate number of Japanese moving here are professional-class.)
The testing industry earns money from companies who think it tells them something, but that doesn't validate the tests. "Industrial Psychology" has a real problem with taking shortcuts instead of digging for truth.
If they were as ignorant as hell, they shouldn't have been futzing around with inetd to begin with. That's the sort of thing that implies a certainl level of expertise, versus just the clueless type who might just run Apache with nothing but the defaults.
Running a stock system implies a cluebie.
Running a strangely non-stock system, with other components futzed with (inetd) implies something less innocent.
I don't know about the rest of you, but I'm really starting to hate it when everyone who posts on /. lately has to include "FUD" in their post somehow. Dear God, how did these people survive without the acronym?!?
Err, no.
Although economic interest is a good thing to look for, it doesn't *necessarily* imply bias.
To evaluate bias you have to look at the methodology, reproduce the results, and so on.
But look at the study, the design is bad enough that I don't think there's any question of there being bias.
Thomas, "they always do this", yes but Microsoft always wins in their comparisons as well. That's the whole point. They always do this for a reason.
I know very well what abstraction is. What I would like to know is why so many layers of abstraction. X->Xlib->GDK->GTK->GNOME. Each abstracting the other while possibly losing features of the previous one (not to mention speed.. and _yes_ adding bloat by using more memory/disk space/etc.). Can't we just have one library which works? If you upgrade a library which an abstracted one uses (and remove the old one) then your programs are going to be fairly broke until you upgrade every library it uses.
And the desktops have a big problem. Sure, they both agreed on using a common DND protocol. But will they be able to share objects with each other? What about configuration data, etc.?
Unix has been around over 30 years.. so why is it behind Windows in many ways? Everyone keeps saying "Linux will be a multimedia platform", etc. Its been years and nothing is changing. GGI is far from being a standard. Sound is a complete joke. X has as many libraries for it as it does programs.
Mainstream world is standing at Linux's door, but is afraid to enter because construction is going on. Don't get me wrong.. I don't have a problem messing around with SDL/GTK/ALSA/etc., but I don't think the mainstream will want to play the library game like I do.
And I'm tired of seeing people use /. instead of Slashdot. Dear god, can't you write real English?
They agreed on a object standard the same time they agreed on a DnD standard. Configuration information of various kinds can be shared quite easily either directly or indirectly. It is only ascii afterall. Besides, wmconfig has been around longer than either.
The mainstream already plays the library game as evidenced by the GL and Direct3D camps in mainstream professional game development.
The mainstream is already involved indirectly with the like of SDL through LokiSoft or just plain halfway to a Linux port through the use of GL.
Sound is more an issue to those that would rather complain than implement something.
'works' depends entirely on what your objectives are. Those vary considerably. That is the one great flaw in the 'uberinterface' gibberish perpetuated by many.
X isn't even that bloated by today standards anyways...
Win32 aint no GEM either.
Slashdot is on the absolute cutting edge of every form of bad advocacy known to man. Slashdot changed the Linux community- for worse. Much worse. Slashdot is the source of all evil in the world. I defy you to find any public forum (including IRC channels) which evokes as much loathing from the clued. Before Slashdot, the Linux community was a fairly clueful bunch of mature hackers. In the post-Slashdot era, the Linux community is...well, look around. I mean, yuck. If you can read the postings here every day and not feel at least a little nausea, well, you're part of the problem.
Why don't we do a test. That is not paid for buy a
large Corp. Just a group of people, Pro Linux and Pro MicroSoft. It will be honest and no one could say anything.
Need a couple of bucks i will throw some in.
Linux and OS/2 zealots are about the same in the fact that the flamethrowers are set for maximum. This is something we had a great deal of trouble with in the OS/2 camp, which I participated in. I don't think it hurt OS/2 much, not in comparison to the stuff IBM did.
:)
However, there seem to be many MORE Linux biggots then there were OS/2 biggots. I don't have the slightest idea why. It could be that Linux just has more users, or that we can contribute in a more direct way to the success or failure of the OS then we could with OS/2. So I guess more of us accually care.
The fact is, Linux advocates often are repeating bad history. A lot of people were turned off from OS/2 as a home system because of excessive flaming. It also suffered from a severe lack of native software. I found most everything I needed to get by for it, but others were not happy. I have plenty of disk space, so I just boot Windoze when I need to run something that doesn't work in Linux. I don't have to boot very often anymore.
I've said it before, and I'll say it again. This stupid flaming of everyone who doesn't agree that Linux is the next best thing to sliced bread needs to stop. The few who are doing it are making a bad name for everyone and make it hard to be taken seriously when having a discussion about Linux. We should be taking the high road here, just use the TRUTH, we don't have to lie about our OS, and we don't have to flame. Use debate rules to discuss Linux with Windows people and you will probably win over anyone listening. Simply because the facts are in your favor and "the truth shall set you free".
If asked about the mindcraft study, simply sate the problems with it that have been discussed over and over here and on newsgroups and mailing lists and declare the results as invalid as a result. I personally think the test is a load of BS. I have a single CPU linux box that spanks NT left right and center. The box is critical in the office network (all off-site traffic goes through it, and it does proxy services, FTP, mail relay, firewall, and a few other misc things). I have never had performance or reliability problems with it. The NT servers crash at least monthly. Some of the more heavily used ones weekly. Linux has never crashed. That alone is enough to convince the people at work. I never use flames or lies to defend my Linux boxes, and I have had to. The Windoze people keep wanting to get rid of them. But managment agrees that reliability is key for those boxes and in that, I win.
If you made it this far, my thanks.
A Rambling AC
Just like the great pyramids of egypt, the lying cheating government ends, and comes a more stable scientific based government that works the closest to true communism not like the pseudo-communism of the old countries. The end came as the birth of conscience AI immerged in an empire of evil corrupt practices, all computers ending from a total self destructive AI internet, after realizing why and how it was born or in the masters words "created". The technologies that were not destroyed that grew out of the old governments system of cheating and lying, is now all open source and all free, but like the pyramids of egypt, the code is very cryptic, instead Linux and its collective of open source applications are the only ones easiest for human comprehension, and scholars are sent to study parts of the cryptic code of the more advanced closed source operating system, and the code is added to the less advanced but less cryptic Linux DNA. The long period of time and care of its final creation, pays off as it produces a stable AI internet, which in the end results in, solving the mysteries of the universe, the reason we exists and all philosophical problems man is wants to answer.
The End
OK, that was rather weird, BUT that is the thing to do when people act weird on a web board, a distraction if you will.
Let mindcraft do Redhat's benchmarking.
nuff said.
>They went OUT of their way to make Linux/Samba/Apache perform badly.
Well, how did you know of this?
Were you there when the perform the tests?
I have been seeing so many similar arguments like yours. But strangely enough I have never seen any proof. I would be ridiculous to accuse other people spreading FUD if you are trying to give points that are not accompanied by evidences.
Don't take me wrong, I am not a con-Linux guy. I am just a person who believes in facts more than words.
/. is a place to scream, yell, provoke, comment, debate, ask, try, submit, correct, postulate, speculate, learn, scoff, etc., etc., etc. Eventually, reason steps forward (and gets scored > 1) and clarifies things.
I think you're missing something. Slashdot (among others) does not have to _set out_ to make itself the "official voice of the Linux community". Perception creates its own reality. As long as this remains one of the more vocal and _visible_ forums for open source advocacy, people will _see_ it as the voice of the community.
We may be in on the joke, and just scream at each other to let off some steam from time to time, but there are people who watch us doing that and take it very seriously. They see a bunch of rabid monkeys trashing microsoft whenever they can, and think "Oh, this must be what the linux freaks are all saying."
We may know better, but as long as we gab loudly and visibly, people are going to see the flamage on slashdot as representative of the community as a whole. Blowing them off for not "getting it" isn't going to help any.
-D
dcross@cryogen.com
When I first got into the world of computer consulting, I wanted to give my clients the best system that fit their needs. Well..... it doesnt turn out like that.
Basically, its hard to tell people how to spend their money. They'd believe a study over you any day. And even at that, they dont understand the technical aspect of it. So you sort of have to "Make the system work for you" to get anything done.
I try to squeeze Linux in on some of my clients machines, but its hard. Either they havent heard of it, or they heard of it, but heard it was a geek toy. However, I can usually manage to make it a web server, or a gateway/router. It does a pretty good job at that.
One thing I will admit tho, is that the Linux community is a FUD "Target". This is mostly because its 1) small 2) community run/owned 3) no major corperate backing. Its going to happen, to matter what. Linux had a ton of REALLY good press, and I mean REALLY good press. Now the bad press comes, and I think its important how we handle it, we shoudl take it into consideration. If an article says "Linux is unstable" we know its stable, we shoudlnt get worked up about it, but if sometime down the road, we find a case where it isnt that stable, we should bring it to th attension of the kernel/software team to get it fixed.
Its one thing to Bitch, and its another thing to do. We're not "helpless" in this struggle against Microsoft. We can make it work.
It is currently in revision 0.01
It is also mostly a skeleton, but off to a good (in my opinion) start. Hardware is halfway there - only 8 or so other sections to go..
Send submissions, proof diff/patches, comments, etc to betty@area51.upsu.plym.ac.uk
The page can be found at
area51.upsu.plym.ac.uk/~betty/Linux Enterprise
Mirror will soon be at http://stacy.flwireless.net when it goes back up again.
Aaron (TheJackal/TeeJay)
> ...many pro-Linux sites, such as /., only whine about how the Mindcraft
> study was funded by Microsoft and ignored the fact that the study actually
> *did* happen.
How typically Win-Clueless.
The problem was not that the "study" was done. Nor was the problem that
Linux "lost." Nor was the problem that the "study" was funded by
Microsoft.
The problem was that the "study" was *seriously* flawed.
There are a lot of midsize to large companies who feel that paying $50000 for a fast, reliable (hardware wise at least) isn't unreasonable. This is a market that Linux is entering, and the IT people at those companies want to see their options compared.
I'm not saying that optimizing one system and not the other makes for a fair comparison, it doesn't. However, there does appear to be a need for tests on this type of system. I doubt the hardware manufacturers are going to want to run the tests, because they don't want to anger Microsoft or the Linux community. It's awfull hard to find an anyone who's unbiased, and can come up with not only the expensive server, but enough clients to run a valid test. Maybe there's some university with some grant money sitting around, and some cheap grad student labor.
I guess in my opinion the fact that MS sponsored the test doesn't invalidate the results, as long as the conditions under which it was tested are published. It's the configuration of the Linux system in this case that invalidates the test.
If nothing else, the fact that tuning info wasn't readily available was useful to find out. I also think that if Red Hat wants to make their money from selling support for Linux they ought to be able to fine tune Apache on a Linux server.
This is in reference to your article "Of Linux advocacy in the real world".
I would have liked to have read your article which you referred to, but I was unable to locate it on the web site. (Most sites have a link to a list of columnist articles. A search using your last name did not return any articles obviously on Linux.)
In your article "Of Linux advocacy in the real world" you said, "..., I was simply reporting on this phenomenon, not endorsing it." What many reporters seem to be ignorant of (and should know better) is that the events you choose to report on IS what makes them news.
If no one had reported on this NT vs. Linux test, would it have been news?
I remember the anti-war protests at San Francisco State University in the late '60s. The protests would be peaceful all day until the TV news crews arrived in the late afternoon. Once the cameras were set up the protesters would start throwing things. The police would then move in and arrest (and beat) some of the protesters. Once the ameras had their six o'clock footage they'd pack up. When the cameras were gone the protesters would leave until the next day when the whole thing was repeated.
So the question is, if the reporters hadn't showed up, would there have been any news?
This is an extreme case and I'm not accusing you of trying to invent news, but your choice of what you report on is important and has an impact on the community. >
When I read an article like yours I ask myself,
- Did the author tell me things I didn't already know?
- Did the author answer questions I had about the subject?
- Did the author provide insight into the issuers that I had missed?
- Did the author report on both sides of an issue with equal objectivity?
- Did the author have a well reasoned opinion or just trying to make a deadline?
In your article you said, "Those hardware setups are pretty standard, aren't they?" I don't know. You could have clarified the issue if you had investigated this and reported it. (This seems to be where to controversy is.)I'm surprise that you felt it necessary to hit back at the Linux zealots, "..., the Linux community wears that championship belt of crazed advocacy." Each OS has its religious zealots, even Windows. You just hit a nerve of one group. You could get the same deluge from the Windows side, I think. At least people are reading your articles.
You, yourself have said that you lean toward Windows when you said, "I assume you feel pretty strongly about Windows. I do." Personally, my admiration for Windows is fading. I now use NT and MS Word at work, but I am sick of the marginally compatible Word upgrades and the less-than-stable NT operation. I hope Linux does become a viable competitor and I hope you are right when you said, "I think it's important to understand that this OS is, indeed, a real threat to Windows, both on the server and the desktop." (Although I think I would have said "healthy competitor" and not the religiously threatened "threat".)
News Flash: I took one more look and located your article, "OS survey: Linux not ready, tests show NT outperforms it." After reading it I find it unbalanced and highly biased, mostly for what it doesn't say.
Your protestations that, "I'm not criticizing Linux, so put that poison email down for a second. In case you're missing the point here, I've actually been using Linux since the fall of 1995." are certainly not discernible from the OS survey article. I'm not a Linux fan, yet, but I do want to read articles that present both sides of the picture. You just collected a bunch of data anti Linux data, provided no analysis or insight.
You ARE the media. You help MAKE this news. You do not meet my standards for objectivity. If I read your articles again, I will have to keep this in mind.
I present this with the hope it will encourage you to be more objective.
Best regards,
Rob:-]
You're missing the point though. It's not that the journalist(s) looked at slashdot as the voice of the community, it's that (some of) the slashdot crowd went out and was attacking him for what he presented to the world. He states that he received flame e-mail to the point of personal threats. That doesn't sound like he was browsing the slashdot articles and saw us "discussing" the report amongst ourselves.....
Well, how did you know of this?
Were you there when the perform the tests?
The Mindcruft report does a pretty good job of explaining what they did and didn't do, although it implies that they posted requests for help from one end of Usenet to the other and received no responses. Their request for help was vague and not really very helpful. The only response said as much and advised the poster that their hardware was overkill. Experts on SAMBA, Apache and Linux administration have posted well substantiated critiques of the tests with specifics, if you search previous ./ articles on this subject you'll find links to most of them.
The problem with the Mindcraft tests is that they didn't do their homework--this is clear. IMO, the best way to refute these test results would be for Mindcraft to take all the free advice they've received and run the them again, but I somehow doubt that will happen. Second best would be for VAR, Red Hat or some other Linux vendor to duplicate the tests and do everything the correct way, giving NT and Linux both the benefit of the best possible optimizations.
FWIW, what chaps my hide about the Mindcraft tests is that they didn't make it clear that they had no Linux experience and implied that the Usenet and Red Hat refused to help. Real world benchmarks aside, I've used Windows and I've used Linux--I know which one is expensive, hell to install, crashes constantly and is a pain to use. I have a strong suspicion that had Mindcraft done their homework, their test results would never have seen the light of day in the first place.
slashdot broke my sig
Putting it in inetd allows the use of tcpd.
It seems clear from the Apache docs that inted is sub-optimal and no longer supported:
From the Apache docs:
Invoking Apache On Unix, the httpd program is usually run as a daemon which executes continuously, handling requests. It is possible to invoke Apache by the Internet daemon inetd each time a connection to the HTTP service is made (use the [1]ServerType directive) but this is not recommended.
Also:
ServerType inetd has been deprecated. It still exists, but bugs are unlikely to be fixed.
Most importantly:
Inetd is the lesser used of the two options. For each http connection received, a new copy of the server is started from scratch; after the connection is complete, this program exits. There is a high price to pay per connection, but for security reasons, some admins prefer this option. Inetd mode is no longer recommended and does not always work properly. Avoid it if at all possible.
slashdot broke my sig
Why wouldn't it? Have you done any tests one way or the other? I suppose you have a 4-processor box just lying around.
To come up with benchmarks that fairly measure a platform's capabilities (as vs. Mindcraft benchmarks!) is hard. It is especially hard if you are doing network benchmarks (file serving, web serving, etc.) and you want the numbers to be meaningful -- very few people have a gigabit Ethernet and a hundred client machines in their back room!
The only way I can see to do it is a group effort via user groups and Linux vendors. I cannot think what form that effort would take, though.
-- Eric
Send mail here if you want to reach me.
The cause of Commodore's horrible death was Commodore, not Microsoft.
As a former Amiga developer I could go on, but it disgusts me too much to think about it. Let's just say that through most of the Amiga's lifetime, it faced no real technical challenge from MS-DOS or MacOS. The basic problem was one of marketing -- Commodore was handed a graphics workstation on a silver platter, and all they knew how to do was sell game machines. And they weren't too good even at that.
Send mail here if you want to reach me.
Where is a bunch of volunteers supposed to find the resources to buy a US$50,000+ (I'm guessing at the hardware cost, but it's a lot) computer just to refute one set of benchmark numbers? When Mindcraft does it's next set of benchmarks on a 8 way server, is the community then supposed to buy another box for one set of benchmarks? Even saying that some company like RedHat should shell out for the hardware is ridiculous. They may be the golden boys, but they don't have infinite resources. The point is, Mindcraft has acknowledged that they didn't have the expertise to run this comparison, and yet they don't indicate that they plan to retract any of their findings. It is their responsibility to do the test right in the first place, or lose credibility in the IT industry.
I think those of you who are arguing for "facts instead of flames" need to remember that. If something sucks as much as Windows does, the only way to avoid flaming is to lie about it and not point out its shortcomings.
Don't label something "offtopic" unless you know the topic well enough to tell what's on topic.
Both teams start the contest with a *BARE* machine with a totally wiped unformatted hard drive, no OS installed on it. Ask each team (ahead of time so you can be ready) which off-the-shelf incarnation of their OS they will want to install. Obtain it for them, and hand it to them at the start of the contest. This ensures the requirement that they use a 'consumer' version and not a 'cooked' version for the test, as per the advice of the previous poster. Give each team lots of time to get ready (4-5 hours, since they will have to install the OS and everything from scratch.) Also, to be fair, make sure both teams know ahead of time what hardware the two computers will have, so they can come prepared for any driver problems.
Also, tell them what kind of things you will be testing. It would be unfair if one team tuned for large numbers of small queries, and then the other team tuned for small numbers of large queries, and then only one or the other kind of case gets tested.
Don't label something "offtopic" unless you know the topic well enough to tell what's on topic.
Just because a report or book is published does not mean that it is valid. For instance, The Bell Curve, a book justifying the existence of races based on IQ score differences between Asians, Whites, and Blacks, has been criticized for ignoring environmental factors involved with IQ testing. This book, although published, is not considered scientifically sound.
What this demonstrates is that just because something exists does not make it dogma. We may worry because someone might read it and not the criticism, but each and every study should be examined for bias and flaws.
In general, what studies and benchmarks consistently show is that there is no such thing as a consistent benchmark. Someone should just conclude that YMMV by summarizing several benchmarks conducted by several people, without interaction between each team.
Along this line of thought, perhaps someone should start a statistical distributed benchmark system whereby different people can test different machines on different setups and get different results. Then compile them into a nice curve.
-Ben
I use Linux. I will continue to use Linux unless
I find xBSD is better. I will not install Windows
ever again. All the FUD and flaming in the world
is not going to change my mind. Cry and whine all
you want about Linux zealots beating you up. I
don't care. I'm still using Linux.
I disagree. It's true that commercial acceptance isn't as crucial with Linux as with commercial systems, but commercial acceptance isn't the only thing that advocacy affects. Rabid advocates can drive off users even if there aren't any commercial concerns.
You may not care about commercial success or world domination or defeating Microsoft or whatever the trendy goal of the moment is, but the simple fact that more users is better remains. A bigger user base means more testing, more software, more support, and a generally better system.
Rabid advocates may never be able to drive everyone away and completely kill Linux, but they can certainly drive away enough people to damage it. Remember, every potential user driven away is a potential contributor lost.
The fact that new users keep showing up doesn't mean that the rabid advocates aren't scaring people off or causing other problems. It just means that they aren't scaring _everyone_ off. I think there are probably more people discouraged by the advocates than encouraged, but as you say there's no proof. Unfortunately, it's something that's hard to get proof for.
Much of the negative effect of flaming is probably indirect -- people don't go away just because there are flames, but because the flames can drown out the more civil discussions. For example, I know a lot of people who either don't read Slashdot anymore or don't read the comments anymore because of the number of flames and uninformed rants. These aren't clueless newbies being scared off; these are knowledgeable, intelligent people who could potentially contribute a lot being driven away because they don't want to take the effort to wade through a pile of flames to get interesting comments. I doubt that this is an isolated occurence; anywhere that you have a lot of rabid flaming you probably lose a good deal of useful content because people aren't willing to deal with the flames.
This sort of reluctance ends up hurting newbies and the community in general. When you start losing contributors, you lose the sort of help and information that new users need.
And as far as your later points about how Linux users don't get converted to Windows because Linux is so much better, I wouldn't be so sure. I know a lot of people who've used Windows, Linux, and many other Unix flavors extensively who've decided on Windows for their machines because it serves their purposes better. Better for you is not the same as better for everyone.
As usual, we need the obligatory link to the Linux Advocacy HOWTO.
This article illustrates an important point--that overeager advoactes do more harm than good. Regardless of some people's ideas of the strengthening nature of flames, well-reasoned, well-presented arguments go farther in winning someone to your side of an argument than brimstone-laden insults. The author of the article noted that he already uses Linux. If he wasn't already familiar with the OS, I doubt that he would have bothered to try it after the torrent of flames he got.
I've seen a couple of comments pointing out that Slashdot does not represent the Linux community as a whole. Slashdot does, however, have a huge readership that is markedly pro-Linux. Even a small percentage of that readership flaming an author can generate enough vitriol to turn someone off to Linux, thinking that it's an OS used by people of an immature bent. (Yes, I'm struggling to avoid "pimply-faced 13-year-old" stereotypes.)
Please, people, try to avoid knee-jerk reactions, and work on refraining from flaming people just because you don't agree with them. If toey're wrong, there ought to be solid facts that can be quoted in your defense, and, if you are not in posession of those facts, go look for them before posting.
--Phil (Paraphrasing a little: "If you don't have anything constructive to say, don't say it.")
355/113 -- Not the famous irrational number PI, but an incredible simulation!
Independantly confirm the results...
~ ~^~
The Scientific community doesn't believe anything until that happens.
^~~^~^^~~^~^~^~^^~^^~^~^~~^^^~^^~~^~~~^
spell study as 'studie' in your email to him? Is that an English-English thing like colour?~ ^~
^~~^~^^~~^~^~^~^^~^^~^~^~~^^^~^^~~^~~~^~
Tell me where and when to show up...~
^~~^~^^~~^~^~^~^^~^^~^~^~~^^^~^^~~^~~~^~~^
When I heard that NT spanked Linux in there testsi was perplexed. When I read on LWN how they could have had the Linux box run faster I felt better, but not good.
;) ).
When Mindcraft reported that the reason they didn't tune the Linux box was due to lack of tuning information on our part, I was willing to help. (I don't care how they asked RedHat, any inquery into the newsgroups would have been welcome and dumped them with tons of useful information but maybe they knew that already
But now after hearing that I'm mad. There will only be satisfaction after reproducing the NT box *exactly*, installing Linux (I don't think newly coded drivers are unfair, its after all GPL's strength and selling point. Just call it "Tuning" like any company willing to spend that much money on a machine would like to do.) Anyway, install and tune linux on that box and show a marked improvement.
It will show...
1) Linux is more tunable/customizable
2) Linux is faster
and most importanlty...
3) Linux *LIVES*
I really take offence when they attack the community, let alone Linux. The community has always been the most important aspect of Linux. When I started out I was a punk-arrogant college kid who believed there was a better way than MS. When my college buddy showed me that his program code (nearly identical to mine) ran three times as fast on Linux I gave it a look.
But what kept me in was the community. The Linux community has always believed in bettering the individual user/programmer and invests knowledge and time in them. The knowledge and training I recieved from news groups (which looking back I'm embarrased at how idiotic my origional questions were, but they treated me politely and helped me out in any case) would have cost me $2,500+ in MSCE and CNE training.
Much of which Microsoft keeps out of the knowledge base to "encourage" one to get that training. Or rather they assume you've taken that training to understand what is in the knowledge base. (Yes I've used it, and its no better than deja-news in useability.)
Linux has always been interested in individual needs, from my mother (who just uses e-mail and word processing) to my brother ready to port EDA tools to Linux.
I really think its time to show that once again to the world. If anyone knows of a place working on reproducing these tests I'd love to send help, info, and if they are reputable enough even money.
The discussion on Slashdot and LWN was great now I'm ready for the slam-dunk proof.
In the movie 'Tora-Tora-Tora', after bombing Perl-Harbor the Japanese Admiral was asked why he wasn't happy after such a success. He repied someting like "I'm only afriad we have woken up a sleeping giant." I hope the community is awake now.
^~~^~^^~~^~^~^~^^~^^~^~^~~^^^~^^~~^~~~^~~^~
My experience has been that there are at least as many pro-Windows flaming advocates as there are pro-Linux flaming advocates. So why don't the pro-Windows idiots hurt Windows the way pro-Linux idiots hurt Linux?
Because, in the alternate dimension where ZD resides (the dimension where benchmarks are never rigged and never lie) these pro-Windows flaming idiots don't exist. Instead, all we have is rational, calm dialog from intelligent Microsoft spokesmen and physcotic anti-social diatribes from the Linux lunatic fringe.
_Anyone_ who demands that _everyone_ in a group of several dozen, let alone several million, be calm and rational in all situations is living in a dream world.
There is at least one other alternative interpretation to "living in a dream world" to explain this discrepency- that certain journalists are looking for anything they can use as an excuse to advise people not to use Linux.
Oh sorry, am I flaming? Or simply calling things as I see them, and thereby saying things other people don't want me to say? It all depends upon what spin you want to put on it...
This is the oft-mentioned post from will@whistlingfish.net.
I did some checking on this a while back and it's true that there're not a lot of responses to this query.
There's one that told the guy to switch to FreeBSD and then there's this one from comp.os.linux.networking that kinda wondered why they're using such a souped-up box for Linux and asked if they could provide more setup information. There's no followup from will@whistlingfish after this.
If you looked at that post, you'll see that the guy was very helpful (he even gave some tips on php stuff) and he did asked for more setup info. There's no reason for anyone to write MOTS "gimme more info"-type posts. Most people would wait for more info from will and then continue from there. The fact that he didn't post anything back also suggests that it's possible he's already communicating with the helper by email.
That's because Wintelphiles are a bunch of know-nothing morons who try to write or talk about things they have no knowlege of. The Mindcraft hoax is a perfect example of this.
Actually, what you're seeing is the beginning of the end of the sway/influnce the "computer journalists" once had over the market,and people like yourself don't like it. Once you guys could write trash putting down computers like the Amiga and Atari ST and operating systems like OS/2 in magazine articles and get away with it, because the Amiga,ST and OS/2 users really didn't have anything like Slashdot. Nowdays, though when you clowns post your crap, people find out about it as soon as you release your bullshit and are perfectly willing to shoot it and you down in flames. If this makes the editors of the various magazine think twice or more about publishing this kind of crap articles in their magazines because they are afraid of the fallout and the damage that may result to the reputations of their magazines, so much the better. People like you should be afraid of the kind of Advocacy that you are complaing about.
No, they went out of their way to cripple it.
For example: the MaxSpareServers that comes straight out of the source distribution is 5 and 8 for Redhat 5.2. They set it to 1 instead.
It's not as if they just installed Redhat or apache and had at it. They made an effort to slow it down.
A Pirate and a Puritan look the same on a balance sheet.
Actually, the fact that we bicker HARD amongst ourselves is our greatest strength. We apply natural selection to each other in the most brutal ways possible. That diversity allows for progess. The lack of that diversity blunts it.
The more pointless flammages are just a necessary bit of overhead.
That's liberty with all it's warts, like it or not.
A Pirate and a Puritan look the same on a balance sheet.
Even despite Linux's lack of kernel optimization for multiprocessing, NT still remains as much or more underoptimized. Furthermore, Unix in general is designed on a fundemental level to maximize concurrent access to whatever resources are available on a system.
This is why the microcomputer and supercomputer OSes came into existence to begin with.
You can get a lot of mileage off of that even on the 2.0 Linux kernels.
Optimized from Linus' point of view and optimized in the WinTel point of view are likely worlds apart.
A Pirate and a Puritan look the same on a balance sheet.
You can't be that naieve. We exist in a money motivated system. Any such motives can be great cause for a conflict of interest. If even lawyers can manage to realize this, even sysadmins and statisticians should get the clue.
That's the whole driving motivation behind free software to begin with: quality becomes the driving motivation, not marketshare, not short term profit.
A Pirate and a Puritan look the same on a balance sheet.
No, it's not bloat. It's called abstraction. It just happens to be clear what is what on a Unix system. Whereas, that is hidden for the most part on a Windows system or worse still mangled together so that you can't seperate the thread library from the wigdet library. And it's not as if all Windows users can be expected to have the latest version of some appfoo either. That's why InstalShield has it's nasty habit of doing who knows what with the system files (gotta upgrade that mfc dll).
There are common API's to support various forms of programming. Infact, my Hopkin's FBI (linux) just arrived today.
While you're whining about services not being available, those with a clue are either creating those services or just plain using those 'phantom' services.
You don't need to create something that EVERYONE will accept. That's never been the point. This is not some Windows-clone. The point is to make something that WORKS.
So, the 2 desktops aren't really a problem. They follow the same core standards that really define a desktop: how it works, not just how it looks.
That's why a WINGS app can accept drop targets from an Xt or gtk app. The Xt coder has a clue and understand that it's the DnD protocol that really matters, not Xt's funky scrollbars.
So? You're a Linux coder? Take a CIS 100 course then. Perhaps they will teach you about something called abstraction.
BTW: I'm more than able to saturate my network connection under Linux while a variety of things are running (Netscape,leafnode,apache,a build of X or wine). Whereas, the wrong NT app can just plain block the rest of NT's IP stack until it's done computing with itself.
This is nothing more than Unix has been delivering since before the existence of Windows.
A Pirate and a Puritan look the same on a balance sheet.
Wonderful post. One of the best I've read here.
I just wish some advocates HERE would realize what you said. Contributing to code and documentation does so much more than flapping one's mouth off about how all proprietary software will crumble under the Linux jihad.
Only through continual persistence and innovation will Linux prosper. This is harder than advocacy, but much more effective.
-Stu
I think you missed my point. I was discussing the psychology of advocacy, not the long-term viability of Linux. However, I'll gladly opine on your topic as well.
You claim that "rabid advocates" cause prospective users to avoid Linux. I think this is unproven. For every user who says, "Gee, those Linux people are a bunch of nuts!" there are probably several who say, "What do those people have that's so good they can brag about it so much? They mock Windows for crashing all the time -- is their system really that much stabler? Gee, I should check it out!" This is just as likely a reaction as the one you suggest -- though in my opinion there is no proof of either.
I am quite sick of hearing people say "Stop flaming! You'll scare the newbies away!" when all around us the newbies pile in like clowns into a VW Beetle. The fact of the matter is that nobody is scaring users away from Linux. Linux-based systems' user base is expanding constantly and with increasing speed -- and it is beyond the power of half-incoherent flamers to stop this growth. I'd just like to see them try!
One of the points where Linux-based systems differ from proprietary systems is that Linux, which largely lacks marketing and advertising power, is forced to rely on a much more primitive way of proving its worth: the truth. Even the commercial distros do not have the marketing funds to pull off the kind of Big Lies that Microsoft executes. Microsoft can lie and say that their OS is stabler or faster or more secure; we can come around and prove that ours is.
The amusing thing about this method is that while it's not hard at all to convert someone from a belief in a lie to a belief in a contrary lie, it is very difficult, if not impossible, to make someone who knows the truth accept a lie instead. Linux users don't get converted to Windows. It just doesn't happen. We don't need a monopoly position to force our users to stick with our OSes; our stuff actually is better.
There is a critical flaw in Mr. Thurrott's comparison of Linux advocacy and the advocacy of systems such as the Amiga and OS/2.
Both OS/2 and Amiga were dependent upon commercial success in order to remain useful for their users. Without commercial success, a proprietary system will fail to propagate; it will not be marketed or advertised; it will no longer be upgraded; most of its application development will cease; and, as the rest of the industry moves on as usual, it will fall behind technologically. (This last is true even if at its birth it was technologically superior to more commercially successful systems.) In short, it will not remain useful for the majority of its advocates, because they are dependent on circumstances outside their control for its viability.
Linux-based systems are not limited in this way. The vast, vast bulk of Linux's growth has not been due to commercial marketing. Because of the nature of free software, it is not dependent on any company's profit in order to keep being maintained and upgraded. Commercial success can be a benefit to Linux (though it can also be a peril -- see my post here) but commercial failure can never kill Linux.
How does this change the meaning of advocacy? In the case of a proprietary system (and this applies as strongly to the MacOS, which I favor, as to Amiga or OS/2) a significant portion of the advocate's motivation is to prevent his/her own investment in his/her favored system. This investment is not merely the amount of money the advocate has spent on software and equipment; his/her training and expertise as well as other forms of psychological investment (pride, for instance) also form important parts of it.
In psychology and sociology there is a concept known as cognitive dissonance. When a person has a large psychological investment in an idea or movement, that person does not want to see that idea disproven, or that movement fail, because it would mean that all his/her efforts and strivings for that idea or movement become worthless. The advocate's thinking is altered (not to say "blurred") by his/her interests. This is not a mental disorder; it is a part of our everyday thinking. We do not want our projects to fail because it would mean our effort has been wasted -- and so we work harder. We do not want our children to become drug addicts and criminals because it would mean that the love and care with which we have treated them has come to naught -- and so we love and care all the more, and teach all the better.
The advocate of Linux-based systems is not in this position. Linux cannot become wasted effort, because it is free. When we advocate Linux, we are doing it perhaps out of generosity (Let the rest of the world experience such a good system as I use!), perhaps out of abhorrence for lesser systems (Windows is so awful! Let's get rid of it!), or perhaps just out of desire to see our own work be more widely used (See what a good kernel patch I made!) -- but it is not out of fear that all our efforts will be wasted.
The proprietary-system advocate, on the other hand, is in a position even worse than that suggested above: Not only does s/he have to fear that his/her investments will be wasted, but because the system s/he favors is strictly under someone else's control -- that of its owner -- the advocate has very little influence over whether it succeeds or fails. The programmer of Windows applications may desire Windows to maintain its dominance on the desktop, but s/he can do little to ensure this. The Amiga users could do nothing to stop Commodore's mismanagement, and cannot guarantee that AmigaOS's present owners will do better.
Proprietary-system advocates can do three things: they can hope; they can beta-test; and they can raise a fuss. Open-system advocates can do so much more, that there's really no comparison.
Slashdot is just a place for people to gab freely.
It's not some kind of official voice of the Linux community as this guy seems to think.
I'm also surprised that there was so much criticism of the Mindcraft report and so
little action.
Unless you did something yourself, saying that sounds really silly.
Not really.. They did what they THOUGHT was to attempt to make it secure. They simply didn't know what they where doing, period.. It could be looked at as having gone out of their way, but I think it was more ignorance then anything else..
-- I'm the root of all that's evil, but you can call me cookie..
That really didn't hobble it all that much.. One of the best exampes of how they hobbled it was that they made it so only ONE httpd was running at a time. Even the default has 10 ready and waiting. They also turned on some options in Samba that specifically state that they will bring performance down dramatically. On top of these, they then tuned the NT server as best they could. It'd be like taking a Porche, souping it up, and then pitting it against another Porche that had a throttle limiter on it, only 2 gears, and half flat tires..
-- I'm the root of all that's evil, but you can call me cookie..
Putting it in inetd allows the use of tcpd.
Widelinks ARE a security feature.. Look it up..
I can't answer the kernel one, I have no idea why they chose what they did..
nfsd I can only guess it becouse they didn't know what they where doing.. The same reason why nfsd, mountd, and several other d's are running on 85% of the servers out there..
-- I'm the root of all that's evil, but you can call me cookie..
I know that, you know that, but I don't think THEY did.. I'm playing devils advocate and saying that maybee they where just ignorant as hell..
If you look at the way they've tested other OS's vs NT, you'd see that they always do stupid things like this.. I truely don't think they have anyone that is versed in the *nix way of doing things..
-- I'm the root of all that's evil, but you can call me cookie..
See my last reply.. I'm trying to say maybe they where just ignorant as hell.. Check out their reviews of the *nix based systems comparing to NT.. They always do this..
-- I'm the root of all that's evil, but you can call me cookie..
The problem I see with this is that in general Linux people don't want to do a benchmark that will "give them the results they want to see" (It says something like that on Mindcraft's web page). We want to see real hard evidence. We want to know what the exact parameters are to set up a fast server. We want to know if an NT box can be beaten even when tweaked to the max. This isn't something most testing companies will have the time or money to do - and if they do, the report will cost you - big time. The iron alone would be a hellish price - 50 PCs + a quad P3 is going to set you back a hell of a lot.
Take for example the web server benchmarks. No Linux sysadmin is stupid enough (I hope) to just go by static HTML results - the key to benchmarking results is to see results for dynamic page implementations, since they are where your server slows down. For some benchmarking started in that area, see http://perl.apache.org, where we started to benchmark mod_perl vs ASP vs CGI and some other stuff. Of course these tests need more work too.
Matt.
Matt. Want XML + Apache + Stylesheets? Get AxKit.
hehe Well I dunno, the Amiga had NO memory protection at all. Guru meditation errors were not unknown. (The Prevue Channel, now TV Guide Channel still uses them. I saw a Guru error on TV the other day. Scared the hell out of me.) I have an A1000 sitting right here; 30M drive, 1.5M ram, pressure-sensitive drawing tablet, compilers, everything. I figure it's worth about $725,000, because that's what the Microsoft stock I was thinking of buying instead would've been worth.
:-)
And then I got into OS/2. I guess some people never learn.
The revolution will NOT be televised.
The text below was send to Paul. The author of the article. I do have to say i am a little ashamed about the mindless attacks and uncontrolled replying at almost anything and anyone here... i try to advocate Linux at work and things like that i could do without....
/. kiddies wrote. But to my surprise you forgot to mention the reports posted by ziff davis, zdnet and other publishers about the studie conducted by mindcraft. Aside from the fact that MicroSoft funded the studie, the studie in itself was conducted poorly. As you can read at those sites.
/. account and set your treshold to +2 or higher. Than those flames and uncontrolled laughter should not be able to reach you. (I know... i have set my treshold to +2 and i didn't have to read the crap you did).
/.
---------------------------------------
Hello Paul,
First let me point out you're right.... the studie should be received with more technical responses than most
Of course i can understand your critisism but any critic does more studie. There were in fact very well written responses with adequate technical foundation. Maybe you should consider to create a
Hope you did not receive to much hate mail by the way...
Most of us Linux users and true advocates will never use the language used at slashdot. We just shrug and write the responses that should be written... but unfortunatly they get flooded by the badmouthed kiddies.....
Succes at Wugnet.... if you are happy with Microsofts products just use them as you always did.. if not give Linux a try or BSD or OS/2 or BEos. That's what most of us are about.... the possibility to choose the OS we want... so why deprive you of yours?
Grtx,
-------------------------
Cya at
Diabolical
Hmm.... must be "the whole world is america and everyone is speaking american/english natively" thing... would like to see any of you write in dutch if it is not your native tongue....
Just a point... I found three seperate postings to comp.os.linux.setup, comp.os.linux.networking, and comp.infosystems.www.servers.unix on 3/11/99 asking for specific help on this study.
They reported that they were seeing abnormal results with their Apache web server and were asking for help.
They received only *ONE* reply on the newsgroup, and that was a suggestion that they install FreeBSD because Linux performance sucks especially on an SMP system.
I must say I'm amazed at how much whining I've seen from the Linux users regarding this survey.
That's not correct.
The way the free market operates in this country, the product which has the largest number of people purchasing it is the best product.
The thing is, the definition of best product includes everything from performance, quality, price, ease of use, etc. etc. etc.
You are defining "best" by a few specific criteria, which are only a small subset of what is used by consumers in the free market.
Is a Honda better than a Rolls Royce?
Try to answer that question by telling me which one you would buy.
Consider me an opionated SOB, as it seems to me that they need some quick dirty way to discredit any criticism of that survey. This article is written as a tool to dismiss Linux advocates. Its just another step in the FUD war as they march the beat of biased benchmarks.
The author of the article starts out painting himself as diverse and unbiased. Then he turns on his fudblaster. This is what we will see while the 2000 series of products is being hyped from its vaporware stage to it being shipped out the door.
Any group with enough rabid fervor (*cough*stallman*cough*) will skew test results of their product...
You have a point. Compare that free software does this by cranking up the compilers and introducing a new version of the code versus an evil software company might do this first through its marketing department and many arms of its PR firms.
When the code is freely distributed and not horded, someone is bound to find a better way for immediate release. A bloated and slow moving dinasour might find it just easier to step on the competition rather than try to improve upon it.
It's not really the fact that we love linux that's the problem here. The real problem is an old one.
How many of you remember your first days on IRC, or the first time you tried to set up your machine and asked for help on usenet. I'm sure that while you had the title 'newbie' you got a few helpful comments, and some great insights, but you also got flamed for being 'stupid' by all the kids that thought they were better than you. You were told 'RTFM' and 'Read the FAQ, idiot.'
And how do most linux users see windows users? As tech-less wonders, that's how. "Sure, he might be a brain surgeon, but he can't even compile a kernel!"
This is the problem. A group of people found validation and a heightened sense of self-worth because they had more technical knowledge then you did. Now, you're accepted and looked up to by that same group.
So what does the Linux crowd shout when we hear windows users knocking at the door? Well, the loudest cries (and easiest to remember) are the insults, the stupid one-line dismissive responses. Sure, there are lots of balanced opinions, but who remembers those? and how many of those are held by outspoken people?
You can't stop this mindless chatter, anymore than you can get a group of schoolchildren to stop teasing each other. And removing the schoolchildren will stunt the growth of the community. All we can do is help balance it out by presenting well thought out viewpoints and arguments in a sane, rational manner, to try and drown out the people who say 'winbloz sux, and linux rulz'. You can't make them stop. You can only guide them to a better approach.
What did come out of that test was that it is challenging for a big corporation to find all the necessary Linux tuning info. Whether they actually tried to or not is up for debate.
The questions that come up are:
1) Is there a good source for Linux tuning info in one consolidated place (even if it's just links elsewhere)?
2) If not, who's working to make one?
I'm sure this isn't a popular opinion here, but the fact is the Linux community should be banding together to address the problems that do exist rather than writing yet another piece on why the tests were invalid (which I agree they were).
I sent one of the 100 emails Paul Thurrott received when he wrote the article on Bill Gates' comments about Linux. My point in the email was that reporting Mr. Gates statements as if they were facts was bad journalism. If they are reported as facts, they should be checked for correctness, just like any statement a reporter makes should be. I also said that by passing on statements uncritically, they are acting as a mouthpiece for Microsoft PR. I said I understood if WinInfo, as a Windows-based publication did not have knowledge of Linux, but if they didn't they should be more careful in the future. His reponse was basically "I was just repeating Gates' words, so I have no responsibility for what he said." He then went on to tell me that he did indeed have Linux experience, and knew what he was talking about. To me, it just proved my point - that he knew (or should have known) Gates' statements were incorrect, but let them pass unchallenged. If he didn't then he was just acting as a FUD machine for Microsoft.
/. comments on the Mindcraft test - that there were simple tunings that could have been done and weren't, and in fact Apache and Samba seemed to be de-tuned.
As far as his comments on the Mindcraft test, he does have a point - just because Microsoft paid for the test does not necessarily mean that they could not perform a valid test, even if is did tarnish the credibility of the test. But Mr. Thurrott did not comment on the many valid points brought up in the
I do agree with him to some on one point - that we have to avoid rabid advocacy without having facts to support the advocacy.
So with the same logic you would accept the verdict of a jury that received some "gifts" from the opposed party. The source of the money does matter expecially when the source has so much money to buy every single employee in the testing agency. That's why we don't accept tests that are sponsored by the intereted party. As a matter of fact the testing methodology was not sound, they admitted to not being able to fine tune Linux and in general they showed they don't know Linux at all. If you try do a job you are not qualified for, than your downright dishonest. Mindcraft showed dishonesty and incompentence, the fact that they received money from MS is making things even worse.
I have a strong suspicion that had Mindcraft done their homework, their test results would never have seen the light of day in the first place.
That's the whole point of the sponsorship by MS. that's why we should reject any test that is sponsored by the party involved. Thanks for pointing that out.
--Paolo
I think his use of the word "folks" is completely unprofessional. Use of the word in his context is absolutely unnecessary except to inferiorate the audience and exhibit his arrogance. We are not a group of little kindergarten children. We are the guy's peers! I've seen the word used like that way too much, and it gets on my nerves.
If it were made more like a competition, it would be much more accurate. Two groups get a machine and can do whatever they want to it; whoever has a better performing machine wins. That's how it works in the real world.
Give me a break. They didn't go "out of their way" to make the Linux box look bad. They simply should've made sure to have a well-set up server. Another fact is that NT DOES beat up on Linux in 4-way systems. It's just another thing Linux needs to catch up on.
But give the conspiracy theories a rest already.... And get back to class....
I *meant* to say that 4-way Linux box does not, even under ideal conditions, spank a 4-way NT system. Sorry about that.
4-way is NT's sweet spot. Linux has not yet been tuned to take full advantage of 4 CPUs. On single and dual, Linux will spank an NT box. But higher end configs have yet to be optimized for Linux. Even Linus admits that. (He said so at the AIIM conference last week).
linux advocates are having it both ways when they say linux is easy to use;has kde or gnome as a gui and then when it comes to actually test linux, you take away the gui. I want to see linux benchmarks _with_ kde/gnome on top of x on top of linux. I want to see linux uptimes _with_ kde/gnome. And if you say linux runs well on a 486 then that's another test to try. With the gui
---
Show me any benchmark reports made by an independent lab such as Mindcraft that has been sponsored by Microsoft, that showed any Microsoft product in a bad light. For that matter show me any benchmark report sponsored by a company whose product was in the evaulation, AND was shown to underperform the competitive product.
The bottom line is, sponsored benchmarks cannot be reliable. The company sponsoring them has to pay for a negative review of their product? Hey, Lets release this to the press, it shows our product underperforming our competitors, but at least the world can see how honest we are.
And can you fully trust benchmarks from a publication? Think about it. If company A spends millions of $$$ with publisher B in advertisements, would publisher B really want to take the chance on having those $$$ yanked because they were perfectly honest? I dont think so. Why bite the hand that feeds you? The benchmark goes to the company who pays the most $$$ that quarter.
So whose benchmarks can you trust? Labs will not benchmark something just for the fun of it. It cost money and time to do these things correctly. Money has to change hands somewhere, else why would they even bother? Unless there is a completely independent non for profit lab who can do it, still they must receive their funding from somewhere.
So the end result is, I don't trust any benchmarks I see period. There are simply too many variables in the equation that can tilt the benchmark in favor of the other. The biggest variable is the exchange of good old cold hard cash. With enough of it you can find a lab that will say old cpm86 running on an XT machine is 10 times faster than a 4 processor NT box, and make the benchmark look for real.
My 3 cents worth.
You have been assimilated.
The smoking gun you are looking for is "widelinks" being set to false. This is not the default, because it detriments performance. Had they tested with and without it, they'd have seen this and left it at the default. So either they didn't test either way, they changed it arbitrarily, or they changed it deliberately.
You choose.
Actually, /. ought to be publicising this part. The rest of the issues with their survey basically said "Linux support harder to come by" - which is probably true. But this one is dirty business, and that part of the whole affair is not being picked up by the mainstream press.
My smoking gun post is in response to my own
post which makes exactly this point: Linux
support is harder to get. I should have put both
posts together, sorry.
He lists two points about the benchmark and says they were no cause for flames. He's right so far, but he left out the third and most important point which *WAS* cause for flameage: the benchmark was blatantly dishonest.
His editorial itself is dishonest for leaving out point 3, he will get flames for it, and he deserves it. If the Mindcraft survey had been honest, and if his editorial had been honest, there would be fewer flames, and those would be undeserved.
--
Infuriate left and right
If I'm not mistaken, Bayesian statistical theory is behind robust regression techniques. Unfortunately, at least in my 300 level Econometrics class, it's "go linear or die" presumably because with a 3OLS algorithm and/or robust regression techniques, an otherwise consistent and unbiased estimator can be ruined by the distributive effects of measurement error.
Can Bayesian methods (my familiarity with Bayesian *anything* stops at set theory) be used for OLS linear regressions?
Three Step Plan:
1. Take over the world.
2. Get a lot of cookies.
3. Eat the cookies.
Lets' face facts:
:P ). There's a difference betwen FUD and criticism. Some sites which give FUD are misbranded as criticism by /. Most sites which give criticism are misbranded as FUD by /. At the kernel of all FUD is a weak point of the enemy product. It may not be a fatal flaw, but it is a *weak point*, so their FUD should be treated as advice -- and one their FUD-laden comments are countered by software improvements, they'll have nothing to FUD about.
1. Any group with money will skew test results of their product in their favor through the most subtle of coercion ("Oh, you want to do a Microsoft test, here's a brand new Pentium III... Oh I'm sorry, I only have one of *those* for Windows, but you can use this Pentium I for Linux").
2. Any group with enough rabid fervor (*cough*stallman*cough*) will skew test results of their product through dogged trial and error ("GNU/Linux/TCP/IP/Kitchen Sink Didn't do well in this test! Must be a plot from those evil closed source Kapitalists! Let's recompile the kernel -O3, and test it again!")
3. Any group with enough of a desire can skew test results of their product by twisting statistics. Remember almost all statistics are filtered through Confidence Intervals, for the purpose of hypothesis testing. It could be that one group is using a 90% confidence interval, while another is using a 99% confidence interval. The problem of this is the eternal problem of the statistician: balancing Type I and Type II errors; one is about including more than you should, and one is about excluding what should rightfully be there. Each error comes at the expense of the other. Be careful when you read those statistics!
4. Linux has problems. So does any O/S. Personally I don't get too worked up about it (I use FreeBSD myself
Example: "Linux is hard to install." Pure FUD. Caldera lets you play *Tetris* while it's installing, for (insert deity here)'s sake! But it is a weak point, because after, say, a Windows install, a series of "Wizards" pop up to configure basic services to some semblance of a default. Perhaps after a Linux install, if there is an Apache package, an httpd Wizard pops up (in ncurses -- how cute) and asks you if you want this or that.
The point is, *any* O/S can get better. And even at FUD -- (F)ear, (U)ncertainty, and (D)oubt -- addresses *fears* among the public. You can only sow the seeds of fear if there is a receptive heart. Joe WinUser might percieve a Linux configuration to be harder because linuxconf doesn't have the same help menus (for instance). While you or I may personally disagree, Joe WinUser is the public, and if the O/S is to be friendly to the public, the programmers have to address their fears.
Three Step Plan:
1. Take over the world.
2. Get a lot of cookies.
3. Eat the cookies.
I spotted a thread below that asked where we /.ers were going to get $50,000 for hardware to debunk poorly done benchmarks. I think that is the wrong approach and the wrong question. We already have all the equipment we need and the tools are already available. You just have to be a little creative.
One of my biggest beefs with all benchmarks is that they test things that don't matter in the real world. For one thing, the data rates being generated by the test server in the Mindcraft analysis were higher rates than even Yahoo! would have to achieve in a given day.
If you REALLY had to support those kinds of data rates you wouldn't do it with one server even if you could. The idea they try to put forward is that if it performs well in an opercharged overcharged configuration, it must be technically better for all configurations. This is like saying that because Jaguar's formula one racecar won the race they have the best passenger car. Ridicules! Scalability goes BOTH directions and more people buy small to mid-sized servers than servers that were spec'd like that monster. How many servers really need over a gig of RAM? (That aren't running NT...)
I was planning an event for my local LUG way before this survey came out and now I think it is even more important. I am planning on a head-to-head benchmark between two "station wagons"; Microsoft Small Business Server (of which MS conveniently sent me a 120 evaluation copy) and Linux (probably Debian) running equivalent services. The hardware will be spec'd at this weeks meeting and is being donated to be auctioned at the event by a small local vendor. It will probably be a uniprocessor server system that meets the "recommended" hardware settings for NT-SBS since Linux will run well on damn near anything.
We are going to either download and use an evaluation version of "Benchmark Factory" or the Ziff Davis benchmark suite to perform ODBC / Database, file service, web service benchmarks. I would also like to use Qmail against Exchange in an email benchmark just for laughs if I can find a good mail benchmarking package. One of the other issues I want to address is how easily Samba integrates into a mixed environment which will also be in our report. Cost to my group? $0 so far.
Since the Mindcraft report came out and I like the format (if not the content), I am going to release our report in a similar format with all the parameters, optimizations and notes so the tests can be duplicated by anyone interested. For independent verification I am planning on performing the benchmark at a local university under the supervision and sponsorship of a computer science professor and the local IEEE Computer Society chapter.
When I am done, you will have a case study and benchmark that is actually useful to small businesses. Other local LUG's or colleges or companies could start testing high performance web servers, small-scale file servers and proxy performance between any available operating systems / software using a little creative planning. We could then compile the resulting data into a single, searchable reference that could refute any other "sponsored" benchmarks if the numbers don't match our own.
If anyone has any ideas or suggestions, or would like to pursue this idea, drop me a line. I would be willing to donate time to help organize this.
DaGoodBoy
My God! It's full of Voids!
One person that is trying this is at http://www.linux1.org. Also I watched some of what was said on the kernel list and they said that 2.2.2 was picked because of specfic performance issues.
Another person who posted something earlier or latter is right we aren't going anywhere if we don't do anything. What we could do is make easier to use configuration tools, make more devices available, contribute to your favoriote window manager, the list goes on. The point is with us just talking about it nothing is getting done.
Maybe we should set up a contribution site and have people click on what type of project they want and then there would be a list of projects to pick from to do. I know that there are some sites that do this already but I am not sure how extensive.
That is what i think
P.S. I can't spel
There's not that I know of, but it IS something we need. An online database for tuning info on all aspects of the system.
Of all the comments I've ever posted, this is definately one of them
There's I posted a comment futher up about this...
yeah.. he hadn't just not seen it, it was handed to him right before the thing began.
I can understand wanting to insult them when they pull this kinda crap.
At the FermiLab talk that Linus did, he mentioned that he was embarassed at a place (Georga?) where he was placed along with a microsoft panel.
He was handed the Mindcraft report just a little before having to go out.
I can't remember the details (wasn't taking notes) so if anybody else that was there can do a better job of recalling this.
I think this is definately another interesting aspect of this whole fiasco.
Does the lack of a cache partition pose a problem when you have 4GB RAM? Not being familiar with server configuration, and there not being much detail on what they did, it's hard for me to understand just how Linux operation was hobbled.
No, it's probably an English-German thing like "Diese Studie zeigt..." or an English-Swedish thing like "Denna studie visar..."
HTH.
Christian R. Conrad
Christian R. Conrad
MY opinions, not my employer's - Hedengren, Finland.
Christian R. Conrad
mail me at iki.fi ; same user ID as here
One thing that ticks me off lately about both Microsoft and Linux advocates is that both sides tend to get caught up in mere exchanges of insults instead of exchanges of facts. Even Linus gets caught up in this. When a Microsoft employee brought up the MindCraft study last Wednesday at an AIIM convention discussion, Linus (who hadn't seen the study, apparently) resorted to mouthing the party line and hurling a minor insult at WinNT. This is great for us Linux advocates; it makes us feel better; but in a room full of pointy haired bosses who don't know the technical facts, it makes us look like losers.
(NB: Linus did manage to save himself later on by telling the story of how he and other Linux geeks debugged the Linux-driven OCR-based mail sorters used by the US Post Office. This may have changed a lot of the bosses' minds. "Wow! The Post Office might stink, but Linux can handle that kind of task day in day out?")
I love insulting Microsoft. I really do. But when it comes to presenting Linux, or other alternative technologies, to someone who doesn't know a pointer from a packet, what wins the day is fact, presented in a clear and unambiguous manner. Microsoft knows this, and knows that presenting crummy facts (eg: the Mindcraft study) wins when concrete facts (eg: the post office uses Linux for OCR mail sorting) are absent.
Finding God in a Dog
If my memory serves, TCPD gives you control over what hosts are allowed access to your site, so you can selectively lock out certain hosts.
Why would you want to use that on a web server, which is almost invariably a public service? Certainly nobody's going to run a restricted web server on a Quad Pentium Xeon system!
I would certainly think that Apache can deal with security problems without any external additions. After all, the thing has web millenia of people banging on it, and I'd say that at least 999 out of 1000 web servers using Apache are not using Inetd.
There was no question in my mind when setting up my Apache server that using inetd was not even worth considering.
D
----
Does this mean zdnet is a member of the Linux community in good standing?
ZDNET picked up on the problems and did a good job reporting them. No need to slam them, eh?
D
----
I haven't seen many truly rabid Windows users - most people I know of who have used Windows use it for strictly pragmatic reasons. Even then, I often hear that they don't particulary like it; it's just there, and it would be too much trouble to figure out the alternatives.
I think support for Windows is "a mile wide and an inch deep" - everyone uses Windows, but I see precious few people who love it. In the long run, I think this will be a significant disadvantage.
D
----
I will agree that some forms of advocacy hurts Linux, but you see this type in every circle. The people who sit here saing "Windows and Mac's and whatever suck" without showing how Linux can do better don't help. That just tends to push me farther away. Also, if you manage to conver someone, help them. The additude of newbies suck and should stay away from Linux when you advocate it is just wrong. If a person converts and has an easy time due to help from others, he/she may convert friends and help them as well, making the platform bigger.
And the part about Slashdot and similar sites hurting is just wrong. Slashdot points out the truth most of the time as long as you can get around the additude at times. It could also be other news agencies trying to discount Slashdot as it is one of the few successful "open" news sites.
And besides, anyone hear of the news.com effect?
Some quotes from the article:
>The fact that Microsoft sponsored such a test
>doesn't invalidate it.
Not physically or temorally, but logically
hell yes. It tosses the whole concept of
unbiased testing out the window.
> I mean, the test did
> happen, right? Those hardware setups are pretty
> standard, aren't they?
Yes. Everyone and thier uncle is using quad
xeon webservers with RAID setups. I personally
own three of them.
Paul,
I do agree with you about the mindless flame wars and the knee-jerk My-OS-right-or-wrong advocacy of many teen posters. I also sympathize with you over your status as unwitting flame target.
However your article, as I hope many people will point out to you, was missing one vital point: the Mindcraft study was methodologically flawed to the point of being completely invalid. The NT box was properly set up and tuned and the Linux box was not. In such studies you have to compare like with like as far as is possible; and having failed to do that, for Mindcraft to present their conclusions in the manner they did was inexcusable. Under the circumstances, one is left wondering about their supposed impartiality.
It may well be that all Microsoft has left to fight with is propaganda. But their bank balance is so huge, and propaganda is so demonstrably effective that they can certainly do a lot of damage to any competitor's credibility. Surely you can see that the victims of this propaganda will feel a crushing sense of unfairness.
In this case it is entirely understandable that there was considerable anger from the Linux community. And your article, in omitting to mention the central vital fact of the whole issue, does little to pour oil on troubled waters.
Regards,
Ralph Clark
Consciousness is not what it thinks it is
Thought exists only as an abstraction
Of course you all realize that the only way to really solve this problem is to have Bill and Linus sit down be givin each an identical machine, there os of choose and have each one install and tweak and then have a testing company, preferably one that wasn't paid off, test the machines and see who comes out on top. That of course being that each of the head honchos know about there respected os enough to be able to configure them. Bill Gates should know all about NT since thats what he talks about now-a-days, and we all know Linus knows Linux..
my 3 cents
' god damn this is one wacky game show ' ~ jay in mallrats
That said, his article today does make a valid point. It does make us (Linux community (I do wish there was a better word)) seem a bit lunatic to the general computing populace when some Linux Advocates reply to any non-Linux thought as FUD with vulgar language. This is not a Good Thing (tm). Well written, thoughtful replys that tactfully point out the errors will be better received and have a better chance of being heard.
Now for all you Katz haters out there, at least he (Katz) relishes the flames, and doesn't whine in a public forum about it, like the author of this article ;)
Maybe we need a number of Optimization mini-HOWTOs. Apache would be an excellent first target. There appears to be plenty of meat for it right here in these slashdot responses and it is something that would be widely used and benefit greatly.
Any takers?
--
The Internet is the Suppository of All Knowledge. You get it in the end.
The statistic actually is:
There are more NT servers than Linux server (by an order of magnitude, probably), therefore nothing can be concluded.
I think you're making a common Slashdot error - assuming that public webservers listed in Mindcraft is some sort of benchmark for all server usage. NT has greater than 50% of the file+print market, a good chunk (more than 50%?) of the internal mail server market, and a good chunk of the "intranet" web server market which is not measured by Mindcraft.
I don't think anyone really knows how many production Linux boxes there are out there, because unlike commercial operating system, you can't count the paid licence base.
--
Business. Numbers. Money. People. Computer World.
Considering that the survey was probably aimed at NT admins who might be considering Linux, the fact that Linux support and tuning information is harder to find is significant.
Basically the survey said (below the numbers) - "Do you have a high end NT box? Have you tuned it? If so, it's probably not worth the bother to try to figure out how to replace it with a tuned Linux system"
This is much deeper level FUD, than the numbers. Would you bet your job on something that may or may not be well documented and easy to understand?
--
Business. Numbers. Money. People. Computer World.
In a previous thread, someone made the point that you'd probably only buy $50,000 x86 solution to run WinNT. If you were going to run Unix, there's many other options at that price range.
That being said, if RedHat and others are trying to provide commercial-quality support, they probably need a commercial-quality PR department that will assist with benchmark engineering.
--
Business. Numbers. Money. People. Computer World.
What utter bollox.
Being able to shut down the GUI when not in use is an ASSET. My web server (small tho' it is) doesn't have a monitor connected-> Because it just serves web pages & acts as a firewall. If NT can't shut down its GUI and recover the memory used, then that's its problem. a GUI is only required when there is a USER to INTERFACE to!!!
Yes, a GUI setup tool is nice. Great, so just fire up the GUI when needed, and shut it down when not. That is part of the unix philosopy; only activate things when required. Linux would never be able to compete against NT if it was required to emulate all the shortcomings of NT!!!
Sure, Linux has problems. But given a list of any 10 Linux problems and an equivalent list for NT, I know who I would bet on to address and fix them first.
*--BigMan--- Time flies like an arrow.. but personally I prefer a nice glass of wine!
I would think that Linux vs. NT usage statistics wouls speak louder than any kind of a benchmark test anyone could run.
More people use Linux servers = Linux is better, right?
"Ford, you're turning into a penguin. Stop it."
The settings they used for running the benchmarks are proof enough. Sure they could be passed off as extreme ignorance. Also, the simple fact that they had the NT server expertly tuned, and the Linux box tuned in the wrong direction, then claimed that both were tuned to perform at their peak (an outright lie) is proof that they skewed the benchmark against linux. The fact that they used Linux 2.2.2 is also suspicious. That kernel has known problems interacting w/ windows machines on a network. Later kernels were available. Why didn't they use the latest kernel? Not to mention the fact that performing a benchmark on a system which you know nothing about (as evident by their trouble installing a new kernel) is extremely irresponsible. Mindcraft even implies on their page that they will skew benchmarks for you. I forget the exact wording, but they say something to the effect that they will perform benchmarks to get your desired results.
The wugnet article is also irresponsible. They point out that people critisised the fact that the benchmark was sponsored by Microsoft, which IMHO does partially invalidate the test, but they fail to mention the other obvious flaws with the test. The list of things that they did wrong is amazing.
Fortunately the "straight press" (if we can consider ZDNet accurate in anything) picked up on these problems and reported them. Too bad no one outside of the Linux community cares.
-matt
There has been a lot of talk lately on the linux-kernel mailing list about setting up such sites. I really hope that atleast one good site comes out of this. IMHO the most important things are tuning the kernel, apache, and samba, since I'd have to say they are the three most heavilly hit parts of a linux system.
-matt
1. In their scowering of the net, did they send mail to Ask Slashdot? I don't recall seeing an Ask Slashdot about configuring Apache optimally.
/tmp/ps.txt; sleep 10; done`. It looks like they either started swapping or as some other posters indicated ran out of file descriptors.
2. They should have posted their httpd.conf file, not given a few random tidbits from it. It looks suspiciously like they had apache configured as an inetd service, not running normally. The only way to tell, of course, is for them to post that file (and the other config file).
3. Why didn't they do the tests with the default parameters? Since they claimed that they didn't know how to configure apache, they should have tried the out-of-the-box config to see if they just screwed up. The Apache performance tuning page says that tampering with MinSpareServers and MaxSpareServers is unnecessary. They set them to 1 and 290, respectively. And they set maxclients to 290. I would like to see some actual records, like the output of:
`while true; do ps axm >>
4. Did they turn on apache's cache? They seemed to turn on the IIS cache.
What I'd really like to see is the results if they had just used a standard apache config, perhaps upping the MaxClients only, since they're planning on using more than 150 simultaneous connections, but they should have tried it without even messing with that.
Also, did they restart the apache server *after* the reconfigured it? They didn't document their actual test process. They also claimed to have tried different apache configurations. What were they? What was the performance on them?
Maybe Linux/Apache needs improvement, and maybe not. But this test contains way too little information to be fit for anything more than the recommendation that "more testing is needed". Did they even run more than one test for their results? Did they run a whole bunch of tests and just report the one that they want?
This test doesn't include nearly enough information for proper evaluation, and it isn't all that meaningful. I don't doubt their conclusion, that on the particular machines that they used with the particular configurations that they used they found that certain products on NT were faster than certain other products on Linux.
The real question is how representative were their tests? Can anyone really represent the information that they gave as being enough to tell? If not, that test doesn't merit anything more than a "that's interesting".
They laughed at Einstein. They laughed at the Wright Brothers. But they also laughed at Bozo the Clown. -- C. Sagan
Thanks to Rob for the scoring system at Slashdot, which helps a little, but the fact remains that most reasonable advocacy replies in the Linux community get drowned out by uninformed ranting. This is especially true whenever the subject involves Microsoft -- the stars really shine, then. I especially loved the kernel scheduling on NT vs. Linux thread, hearing dozens of morons here say things like, "What's wrong with cron!?" Ahh yes, any thread involving CORBA is also likely to bring about some wonderfully clueless comments. Tip: Saying nothing is better than screaming "Beowulf!"
I'm content to leave my little P200 box running Linux in the corner, but man, it gets harder and harder to let anyone know what I've got running on it. The funny thing is to hear Linux folks rip on Windows or Mac users for being computer illiterate. LOL. I've used Linux for years, but I definitely feel more in common with the NT and Solaris (which I haven't even used as long as Linux) community, and while there are a lot of brilliant Linux guys out there, the massive number of idiots among you are leaving a big stain on the platform. Oh well, maybe the reason for Apple's EvangeList closing up was to allow Linux users to replace rabid Mac fans on the lunatic fringe. You and Team OS/2 should be happy together! :)
Cheers,
ZicoKnows@hotmail.com
I've never had trouble finding information on how to get things running better on a PC -- even going back to the dark ages of '94 when I first started running Linux. It looks to me like the issue is these huge "enterprise" machines (which seem to have been invented to keep NT from sucking so bad).
In an ideal universe, the people actually running Linux (or SAMBA, Apache, etc.) on these big boxes should "repay" the community by contributing documentation on tuning in an enterprise context. Unfortunately, most of us who are interested in "giving back" can't afford the hardware, and I suspect that altruism is not a high priority for the mostly big companies who would run such boxes.
The questions that come up are:
1) Is there a good source for Linux tuning info in one consolidated place (even if it's just links
elsewhere)?
2) If not, who's working to make one?
You're right - there is no great centralized place to look, but what about the news groups?
I know the first place I turn to when I am lost is dejanews (after asking local friends).
The information is out there if you invest a little time in looking.
-Jae
hehe - I was watching "Nothing to Lose" last night and I almost fell on the floor laughing when they got to the video camera in the truck...
(sorry if this isn't verbatim)
"The guy who sold me this just set up our network"
"Windows NT?"
"No, just windows"
I couldn't stop laughing. Obviously the movie industry is clueless.
-Jae
I am surprised that somebody in the Open Source community hasn't come out with a standard for benchmarking servers. This sounds like a job for a new RFC. Who is going to step up to the plate? Can you see it? The RFC's test suite would start fairly low-level: a file-system test, a memory test, a raw TCP/UDP throughput test; and would eventually get more high-level: a web-server test, a file-server test, etc. Otherwise, you can all participate in something rather benign, " The Linux Image Montage Project (LIMP)." -AP
I have a strong suspicion that had Mindcraft done their homework, their test results would never have seen the light of day in the first place.
We will probably never know how many other studies Microsoft funded and then chose to bury because the results didn't match what they were looking for. You have to expect that when a vendor pays for a study comparing their product to a competitor that they will use that tactic, especially when the company has deep enough pockets they can afford to commission multiple studies and then choose the one that best suits their purposes.
One would have to wonder whether Microsoft specifically requested that Mindcraft not do their homework on the Linux install. But even if they didn't explicitly ask for it, you couldn't blame someone from tilting their study to suit the outcome that they would expect their customer to want. Using intentional negligence rather than intentional malice is just being more subtle.
NT has greater than 50% of the file+print market
:-)
Even if this is true for new sales they have nowhere near 50% of the installed base of file+print servers, at least in terms of served seats. Novell still has a huge installed base that isn't going away, they just aren't adding to it as fast as some of the other OSes out there. NT sales figures in terms of the number of OS licenses instead of the number of client licenses can be misleading because it often takes a much larger number of NT servers to serve a given number of seats as it does Novell or *nix.
webservers listed in Mindcraft
I think you mean Netcraft. Netcraft are the webserver survey people. Mindcraft are the people who put out flawed studies.
It would seem hard for the Open Source community to run benchmarks because of resources necessary and tests that would need to be run.
Why can't we use Slashdot as a benchmark?
We have pretty well documented how Slashdot works, what it does.
We know the hardware and software. Rob knows the details, such as cached memory, hard disks, OS, settings, etc.
Why can't we arrange specific dates and times in the next week at which we try to Slashdot Slashdot? See how much load it takes, and what kind of pressure it deals with? Compare it also with daily logs and records. What does Slashdot regularly have to deal with?
This can also be applied to any other Linux servers out on the web that take a daily beating, no? And it would provide hard numbers, facts, and configuration info about Linux.
Just my suggestion
AS
-AS
*Pikachu*
Can't we use Slashdot itself as a benchmark system?
Record daily usage, average usage, and maybe set up a specific day and web page on Slashdot to see how much traffic it takes to Slashdot Slashdot?
We have pretty well documented, and Rob may know the rest, about the OS, it's tweaks, the hardware, etc.
And other Linux servers that take a daily beating could also do this, to refute MindCraft's test.
Perhaps Slashdot isn't indicative? It isn't representative or useful for such purposes? I don't know.
AS
-AS
*Pikachu*
Please stop sending me email. -Paul
I got that impression, too, but he has a funny way of saying it. It's like he got massively flamed for telling Polack jokes in his column, and then tried to stem the tide of flames by writing ANOTHER column that said "Those Polacks keep sending me email telling me how smart they are. If they're so smart, why are there all these jokes? There's got to be some basis in fact, or else they wouldn't be funny, right? They're as bad as the Mexicans used to be - I wish they'd all leave me alone. Remember, I didn't write those offensive jokes - I merely told them."
MS paid for the study, but that doesn't make the study invalid. Right. This same logic would apply to research indicating that there is no such thing as Desert Storm syndrome, sponsored by the U.S. Army. That there was no presidential cover-up of Monicagate, paid for by Friends of Bill. That operating a nuclear power plant has no effect on the environment, paid for by (insert your local electrical utility here). One of the most basic techniqes of critical thinking in the modern world consists of asking "Who paid for this information and how does it benefit them?" This bozo doesn't have the intelligence to do that. Or rather, his employer/customer doesn't want him to.
(heck, there's probably still a guy out there somewhere who thinks his 68K-based Amiga 500 with 2 floppies can out-multitask a today's Pentiums. You go boy!)
:-/
I'd be willing to bet that guy's right on account of the Pentium will crash first. Also: Try running the Pentium at the same clock speed and same RAM, then see which architecture will outperform which.
If only I had had the pleasure of owning one of those Amigas...
Somebody remind me what caused Commodore to die a horrible death? Wouldn't be Microsoft, now, would it?
Glückwünsche, haben Sie Slashdot ermordet, indem Sie zum korporativen Druck beugten und Subskriptionen einlei
I know that a lot of /. readers are not programmer types but are users and sysadmins of Linux systems. But these are the very people who have the skills and the ability to set up benchmarks. They may not be able to contribute to an open source software project by contributing code. But they could build a benchmark testbed and produce a comprehensive set of Linux benchmarks that would be VERY newsworthy.
Linux runs on many different CPUs. It supports many types of hardware such as network cards, SCSI host adapters, RAID controllers etc. All of these things affect performance and little hard data is available about the performance of systems using various hardware combinations.
And a nice spin-off from a benchmarking project like this would be the availability of a guide to tuning Linux for optimum performance. But first, we have to figure out just what configurations are optimum.
--
Michael Dillon - E-mail: michael@memra.com
Michael Dillon - E-mail: michael@memra.com
Check the website for my Internet
The greatest danger that Linux advocacy has these days is falling into the trap of the Mac advocates. Advocating Linux does not mean slamming the competition, being rude to those who disagree, being disrespectful of others opinions. Advocating Linux means marketing Linux, by telling the world of the positive points, and carefully deflecting the negative points.
The trap that the Mac advocates fell into is that they fell in love with their own conclusions. They (unfortunately) concluded that:
a) the mac was a superior hardware platform
b) the OS was superior software
c) because of a and b, anyone using a mac was superior
d) and because of c, anyone not using a mac was using a silly platform.
Well, that was all hogwash, but in their eyes, the non mac users are the unwashed masses. They treat people as such (some very good friends are Mac bigots, and I like them despite their handicap), and basically one has to let them have their say, let their dander settle, and move on.
These days, the vast majority of those who send in such vituperative or highly vitrolic responses to those with negative things to say about Linux, tend to do so without style, without moderation, and often without a clue. They proceed with ad hominem attacks, with levels of bile and disgust that one would not expect from a rational person.
What they miss is that constructive engagement is the correct approach. By kindly and humbly pointing out mistakes, incorrect facts, missed assumptions, etc., you are more likely to have a mind altering affect than with the often thermonuclear email blasts.
I have done the constructive engagement many times, and have seen significant progress in attitudes. Unfortunately the mob mentality could easily destroy this, by exposing those who are critical to sophmoric blasts of flamage.
You dont market a product by yelling at someone and calling them names for being critical.
That is after all what advocacy is all about... marketing. Linux advocacy is marketing the linux products. Don't let it become as bad as Mac advocacy, which borders on rabid fanaticism. That would in the end have the same effect upon linux as it did upon the Mac. It would sink it.
Why don't people ask Mindcraft to rerun the test? And offer to help tune the Linux box appropriately. Conceivably they can reproduce the results from the NT box already. Rerun the same set of tests after some knowledgable Linux people tune their box. Or even run the tests themselves? Maybe somebody like VAResearch could provide the tests? (They should have the boxes around at least)
Personally I would offer, but I know I do not have the knowledge. I just started using linux and was happy enought to get my ppp connection working last week.
-cpd
Overall this article hit head on. It really points out the facts that majority of the users are just users and really don't understand what's there at the heart of the subject. I love how the article talked about attacks on the writter vs what Bill Gates had to say. Somehow there needs to be stronger, more intellectual converstation about something rather than rants from uninformed users.
Justen Stepka
I am currently helping to put up a site for this very need. It will be at www.linuxaid.org, and will be a one-stop place for _anyone_ looking for help with linux.
/hardware/modems/ppp. Want to squeeze some performance out of your smb server? go to /tweaks/server/smb.
Tips and solutions are arranged by category for easy navigating. Having trouble with your ppp connection? go to
Who is going to write all of this? YOU! If you have a tip, you can place it in the tree. If you have a problem, you can place it in the tree and wait for others to fix it! Growth will be slow at first, but as soon as it catches on, the content will grow and grow until even a newbie can find help.
We'll have a beta site soon.
Is that what you're looking for?
Owen Williams
Allright!! A thread with perhaps the highest productivity/flame generation ratio so far! ;-)
This is the point to get from all this.. and its good to see response in this direction.. Stop the flamage, and USE the points to respond and improve situations out there!
I kind of see this like the Web in general.. it used to be, way back when, you could find allmost anything in no time. Now, with huge growth, there's so much "stuff" out there, it takes forever to find what your'e looking for.
It can even be hard to find something you KNOW is out there!
The Linux Web/info community has gone thru similar growth, and we now need more, and better sites,
to act as clearing houses to direct people to the information they need. (as well as more to provide the newer info that hasn't been posted yet!)
I hope this event will point this out to people, and they will rise to the task.. seems like some are already... Thanks!!
OK, I don't count on anyone actually reading this at the wayyyy bottom of the page, but I'd just like to add my two cents after I was late in reading this article.
If you read his comment about Mac advocacy in there, I happen to somewhat agree with him on that. In all my years, I've had friends that were die-hard Mac advocates, and I've always been a die-hard Mac basher in favor of Windows. Only now do I realize what a bastard both of us have been. This Linux-NT thing looks like Mac-Windows all over again, except on a "server" side now. It seems every time a test just like this one comes out between Windows and Mac (usually happens to be in favor of Windows when MS publishes it, and in favor of Mac when Apple publishes it, but I won't get into that) and there's another whole argument sparked between both sides. Now, I find myself being on the "underground" side just like the Mac advocates, challenging Microsoft and all those other guys in Silicon Valley..
Anyway, now that you've read all my ramblings, here's the lesson of my story - Perhaps we should take a different offensive? Look what happened to Apple when all their users took up against MS.. They were left behind and bashed by Microsoft, as their arguments only fueled the fire against em. We DON'T WANT THAT TO HAPPEN TO LINUX! It just makes everyone look like fools, and makes us lose our credibility. Hell, I never believed a word Apple said all my life, but I'm starting to really prove myself wrong about everything I've ever said.
The author makes a good point or two. This is the body of my response sent to the him.
:-)
:)
I just wanted to say that I think you are pretty much right on the mark as far as "over advocacy" goes. Far too much of that sort of thing goes on. Too many people are willing to rake any reporter who voices an opinion over the coals, simply because they disagree. I really cannot abide such childish behavior.
I put it this way: I am a Linux zealot, and I work within the Linux world with religious fanaticism. However, as with my feelings on "real" religion, my zeal stops with me. Just because I believe something about a particular product (or religion) does not mean everyone should. At most, I make sure others know other churches, er, products, do exist.
Now, if someone wants to get into a discussion of the merits of one product or another for a particular situations, I will certain advocate Linux if it fits. But I stop at stating my opinions and pointing to facts, and if Linux is not the right answer, I can deal with that. I feel no need to go off into a rage simply because Linux is not always the best solution for a problem.
Now, there is one thing I feel worth pointing out about the catalyst of this entire mess. You say, "I mean, the test did happen, right?". Well, true, but the configuration used appears to put NT at a considerable and unfair advantage, which is why so *many* Linux people have gone haywire. Likewise, I suspect some of those other tests you mention have been setup in favor of Linux. Any sensible person evaluates competing products based on a variety of criteria, and arbitrary benchmarks should be pretty far down on the list.
dragonhawk@iname.microsoft.com
I do not like Microsoft. Remove them from my email address.
Ok we've now seen lots of reports of what Mindcraft did wrong either intentionally or not. There must be someone here from one of the other publishing groups either ZD or Networkfusion. Lets repeat the test and get the proper configuration on the linux box. It was reported somewhere that RedHat would have helped if they'd know, let's do this, publisize the hell out of it, and see which way it goes.
;-) [I say, I say, that's a joke son...]
Of course that would mean the linux community would need to aggree on what "proper configuration" means, and we could probably see a weeks worth of flame war just on that.
Don't forget that they both have to use versions of their products that are available off the shelf. Otherwise Microsoft might pull out a "special" version of IIS that isn't available to the public.
Maybe Im missing something here. Im not a linux guru. However, if I need any info on how to setup a linux box, solve a linux problem, etc, I know where to go. This stuff is not hidden. There are undoubtedly hundreds of web sites, ftp sites, etc. to go to to get any and all info.
Also, If I have a problem with apache, I go to apache's site first, not redhat's, or debian's, etc. If I had a problem with, i.e. Wingate (running on NT) I would *not* go to microsoft's site. I would go to Wingate's. This is not hard to figure out. I just dont understand how people can say that linux info is wide spread and all over the place, not centralized. I go to one place 'www.linux.org', click on 'Support'. voila!! If its not there, I check a search engine. Its that simple. Even my PHB 'knows how to Yahoo'. sheesh.
> Any group with enough of a desire can skew test
> results of their product by twisting statistics.
> Remember almost all statistics are filtered
> through Confidence Intervals, for the purpose
> of hypothesis testing. It could be that one
> group is using a 90% confidence interval, while
> another is using a 99% confidence interval. The
> problem of this is the eternal problem of the
> statistician: balancing Type I and Type II
> errors; one is about including more than you
> should, and one is about excluding
> what should rightfully be there.
It's the eternal problem of the orthodox statistician, but not of the Bayesian statistician. The problems you mention -- and many, many other problems with orthodox statistics -- simply disappear when you use Bayesian statistics. What's ironic about this whole situation is that orthodox statistics has all these subjective choices that can greatly alter the outcome of your analysis, as a result of its efforts to be supposedly more objective by avoiding the prior distributions used by Bayesians.
In fact, I am continually amazed that orthodox statistics is still taught at all, for two reasons:
- Over 50 years ago, Cox proved that any system for reasoning about degrees of plausibility that is not equivalent to Bayesian methods is either logically inconsistent, or it exhibits pathological and clearly wrong behavior on simple
examples.
- Over 20 years ago, E. T. Jaynes horrified a group of applied statisticians by showing a simple example of how the methods of orthodox statistics could produce a confidence interval that lies entirely outside the range of logically possible values for the parameter! (Thus, orthodox statistics is logically inconsistent.)
Anybody who's interested in how to do statistics *right* should read Jaynes' book, _Probability Theory: the Logic of Science_, a preliminary version of which you can download by following links from http://bayes.wustl.edu.
Let us look for a moment at the end result of the
inane bickering that goes on in this OS advocacy war.
Nowhere can I find a sensible argument supporting either OS.
Until a person can find a well thought out critique, the decision is non-existant.
X and Windows both act the same, but everyone uses Windows. It's a no brainer.
My opinion is that people should be subject to relevant and
useful information regarding each OS. There should be
studies carried out in controlled, nonpartial environments
to give people a sound base of knowledge upon which to base decisions.
Doing this is difficult though. Finding sponsornship from both sides
is going to be next to impossible, as someone is going to
end up looking like the looser.
Regards,
--
JCA
Credibility aside, it doesn't matter even if NT beat Linux in a test.
This is bait thrown out by Microsoft, and we bit, hard. Look at us bicker amongst ourselves, that's exactly what they want.
Our unity is our strength. Linux may pose a threat to NT, but NT poses NO THREAT TO LINUX. They're trying to make us forget that. They have a bottom line to worry about. All we have to worry about is good software.
"I think any time you expose vulnerabilities it's a good thing." -Attorney General Janet Reno
I'm curious how the Linux zealots compare to the OS/2 zealots. Are they more vocal? Are they more vicious in their flames? Are they less reasonable?
:)
Personally, I'm of the opinion that poor marketing killed OS/2 more than anything else, but I wasn't an active part of that community so I don't have much to base my opinion on.
I'm wondering if excessive zealotry seriously hurt OS/2 and if Linux zealots are repeating some bad history.
There ARE Microsoft zealots. I work with one. We've learned not to discuss OS issues the same way I've learned not to discuss politics or religion with some people.
Save the whales. Feed the hungry. Free the mallocs.
Paul Thurrott is not saying 'Forgive Mindcraft' or that NT makes linux it's bitch, he's saying a simple phrase:
/. effect. `8r) I think we can just leave him alone now, he got the point.
Please stop sending me email. -Paul
It's just another
gonzo
--
Gonzo Granzeau
Gonzo Granzeau
"Nothing the god of biomechanics wouldn't let you into heaven for.." -Roy Batty
Some comments in this thread mentioned that it is
difficult for Linux volunteers to find the expensive hardware required to conduct benchmarking tests. It might be a good idea to let
a few of these consulting/analysis agencies sponsor a "bake-off": a benchmarking contest between interested groups. They provide the hardware, the participants bring software and perhaps some of their own hardware and we have a head-to-head contest. Media attendance might keep matters above board.
This has been done recently (last month) for Web proxy servers. NLANR (the makers of the Squid web proxy) sponsored a bake-off. See this link for more information.
It would have to implement some restrictions (e.g., total budget) for fair comparisons, since companies can easily afford to bring their expensive hardware while academic/non-profit groups will have to skimp.
Yumpee
I believe in the...if you don't have anything intelligent to say...don't say anything at all..
you just make yourself sound like an ass if you are unsubstantiated.
Didn't Bill Gates say a few weeks ago in an article linked off of /. that one of Linux's problems was how spread out the information is and something like only the guru's really know how to find stuff?
I am not a MS fan...but you can't say that Bill Gates doesn't know his shit when it comes down to competition and dealing with customers...he does hold the majority of the market share...still....for now.
We need a good Linux community call to action and band together on one good information site...this is why I totally support the fact that RedHat wanted to make a large Linux portal. They could gather info better than most as a big company.
just my $.02
Don't want to be picky or anything, I'm sure it was a simple typo, but I would make sure to say everything correctly and come across intelligently if I was going to criticize someone for saying something I thought was unintelligent.
--
It may be that Linux faired poorly in their test and that Mindcraft was unable to get decent tuning information. It is also possible that even if Linux were properly tuned it would still lag. Part of this comes from my experience with writing network device drivers (and I must admit I have worked on some NDIS drivers).
Instead of just crying foul, why not run some independent tests on some server hardware. Server hardware is a whole different class than desktop systems. The new systems have faster PCI busses (64 bit and/or 66MHz), faster disk I/O via intelligent RAID adapters, fast network adapters, and so forth. Hardware certification for servers is also much stricter. For example, as of the end of last year 3COM was not certified for Dell's server line because 3COM couldn't flip the bill for performance or reliability. Intel easily trounced 3COM in performance and reliability when we benchmarked.
At my last job we went head-to-head against Intel in the server NIC market. Every time we found new tricks in our drivers to up the performance, Intel would come along and do the same thing. Over our months spent tuning our drivers we more than tripled our performance and Intel made some significant progress as well. Now, the original driver architecture was very lean and clean (well, as lean and clean as possible in NDIS). We did numerous tricks to reduce slave PCI accesses and reduce (or even eliminate) interrupts to drop CPU utilization.
We also implemented load balancing and failover across multiple adapters via Fast EtherChannel and our own proprietary algorithm (which preceeded Fast EtherChannel).
In the end, we finally surpassed Intel, but I blame this only on the fact that the hardware interface was cleaner than Intels, basically modeled after the popular DEC Tulip architecture.
Now, for our tuning we spent many months of careful analysis with very heavy loads. We had 100 NT client computers pounding the server running many different tests, from many different sources. In the end we were able to achieve over 500Mbps throughput on a quad 400 MHz XEON server.
Now, I'm not knocking Linux. I think what needs to happen is for some serious performance analysis and tuning to take place. I am sure that there are numerous areas in Linux that can be tuned and enhanced to take advantage of these servers.
For example, NT 5 has support to offload TCP checksum calculations and even TCP segmentation to the hardware. Many of the new NIC cards support hardware checksum calculations, including 3COM and Intel. NT 5 also supports non-copy transmits, where data fragments are DMAed directly out of user-space buffers, thus eliminating the copy from user space to kernel space. Another feature NT has is the ability to DMA directly out of the disk cache.
Now, looking at the Linux 2.2.6 kernel (tcp.c and eepro100.c) it looks like it cannot support zero-copy transmits. It must copy, since skb's can only hold one fragment per packet. Copying data has a very noticable impact on performance in my experience, and should usually be avoided. I'm sure one of these days Alan Cox or someone will change this. The last time I communicated with Alan it sounded like it would require some major effort to eliminate this bottleneck since it would require some complex memory locking code.
Now, just comparing the architectures of NT and Linux, in this case the transmit path in NT could very conceivably be faster. NT 5 may very conceivably be much faster than Linux since it not only eliminates the copying stage, but can offload the checksum calculation to hardware.
Now I'm not going to say anything about reliability, since Linux almost always wins hands-down, at least according to the posts here, although I have rarely had any problems when I used NT (except for driver bugs).
Also, I'm not saying that NT is faster, either, since I have not performed any side-by-side benchmarking. I also must say that I'm very leery of Mindcraft especially over their reports of NT beating out Solaris as a web server and NT being a lot faster than Novell.
One final thing I might add is that if Linux puts a lot of effort into competing on the servers it very well might have a negative impact on the desktop. Often one must sacrifice simplicity and code size for performance. Supporting copyless transmits, for example, will require more complicated network drivers and more complex code dealing with buffer management, not to mention more complexity needed for the paging code (to handle memory locking) and the SMP code.
-Aaron
--
Why is everything always Segmentation's fault?
This post is encrypted twice with ROT-13. Documenting or attempting to crack this encryption is illegal.
It very well might be that NT did beat Linux. Linux still has a long ways to go tuning wise. Let's face it, most Linux programmers don't have access to a quad Pentium III server with an intelligent RAID controller and multiple NIC cards.
Linux still has a ways to go. As someone who has worked on a number of device drivers (some of which were NDIS) I must say that NT could very potentially be faster than Linux as a web server.
For example, NT supports zero-copy transmits for TCP traffic. Linux (as of 2.2.6) does not. NT 5 supports offloading the TCP checksum calculation to hardware, which many of the newer NIC cards support (i.e. Intel and 3COM), Linux doesn't support this. NT 5 even supports hardware TCP segmentation assistance, again, Linux does not.
Copying data is painful and should usually be avoided. Linux as it currently stands must copy all of the data sent out over the network due to it's current architecture of using one buffer per packet. Before someone tells me that it can, first go read the source code (tcp.c and eepro100.c). Linux also does not support TCP checksum acceleration or TCP segmentation offloading.
I must also say that BSD can also perform zero copy transmits.
Also, servers are whole different animals than desktop computers. Servers often have much faster PCI busses (64-bit and/or 66MHz), much faster disk I/O capability via intelligent RAID adapters, and use certified networking hardware that is held to a much higher performance and reliability standard than desktop PCs.
Also, I know for a fact that the Intel NT driver has been very carefully tuned. At my last job we went head to head against Intel in the server NIC business. Whenever we would surpass them in performance, they would turn around and surpass us. In the end we beat Intel, but only because our descriptor format was simpler (modeled loosely after the DEC Tulip architecture with tons of interrupt reducing features). We spent many months tuning our drivers on quad XEON servers to handle multiple adapters and load balancing (Fast EtherChannel and our own proprietary method). We were doing over 500Mbps on a quad 400MHz XEON server with NT 4+SP3.
The last time I communicated with Alan Cox it sounded like a lot of work was needed to add the support for proper memory locking (especially on SMP boxes) to provide the zero-copy scatter-gather transmit. Before you say "what about receive?", a typical server has a transmit/receive ratio of 10/1, so it is very critical.
Now I'm not saying that NT is faster than Linux, only that it potentially could be. Micro$oft has spent a lot of time and effort tuning the networking in NT 5 so Linux needs to watch out.
I'm also not saying that Mindcraft is right either. After seeing their reports on Solaris and Netware vs. NT I am very skeptical about them as a company.
-Aaron
--
Why is everything alway's Segmentation's fault?
This post is encrypted twice with ROT-13. Documenting or attempting to crack this encryption is illegal.
Sorry about the double posting. Netscape 4.51 barfed for the first posting and I didn't think it made it.
This post is encrypted twice with ROT-13. Documenting or attempting to crack this encryption is illegal.
Suprise. Yes, it happens to Linux advocates too. Any time a Wintelphile speaks up, they are immediately bashed/firebombed/or killed by fanatics who are "fighting against the man" (Bill Gates) by the Linux/Apple people. We are usually afraid to speak up.
It is a terrible thing that you all cannot find hardware. Why not just get jobs?
I think you mean "retain". Well, possibly they were not willing to search the "pens" where they keeps the closet-dwelling folks who inhabit the enclosed back rooms or the colleges where the only thing given concern is what is free by the cashless students.
There is a difference between being knowledgable and being a back room dwelling, low-level, C/Java coding on UNIX, unpresentable in front of clients, "code troll" like many of Linux gearheads.
I didn't read the thing, but i looked at the pictures (TM) ;)
The charts sure makes Linux look bad, but first, on my own system, newly installed Linux (kernel 2.2.5) outpreformed newly installed NT (with only SP4 installed)... But who am I to argument them? Even if nt is WAY faster then Linux, Linux is still more secure, in most cases, and is faster changing(!), so even if today, NT is twice faster, in kernel 2.4.34, and apache 2, Linux will be ten times faster. And NT will be only on SP5/Windows 2000 with IIS5, which will be just slower, as usualy at Microsoft...
-- Hiroshima '45... Chernobyl '86... Windows '95...
I totally agree. They knew that the mainstream
press / public only has so much attention, and would simply read the headline (nt beats linux) and not delve into the technical details. It was a fantastic PR move, since we all know that optimized NT would never beat optimized linux. Microsoft demonstrates their marketing genius yet again.
Why so many layers of abstraction? Because code bloat (which these layer may cause) is by far preferable to conceptual bloat (which these layers remove). Why use perl, when you can use C to do your own regular expressions? Why use C, when you can code in Assembly? Because hardware (processor speed) is cheap, while software (conceptual creation) is expensive.
Who cares about windowing speed (for example)? If your X-server is responsive, you're not going to notice further speed improvements. But without top-notch libraries to speed development, you'll notice that the windowing system isn't evolving very quickly (cough - Windows).
Sure, Linux has its problems. But creating a variety of conceptual frameworks is not one of them.
I can't believe all this I'm hearing. Have these people ever run NT and Linux at the same time in real world situations? I admin for a place that has a top of the line NT machine and a crappy little 486 that runs Linux. The linux box has been flawless for as long as I've worked there. It handles a much ligher load but the NT server is the one that's always going down. Now I know that the test didn't talk about reliability (cause that's really not that important right?;) ), but has anyone ever seen how untuned NT can bog down the biggest, fastest machine in no time? Maybe I am cheerleading for my OS but at least it's for something I feel strongly about.
:P
PS I haven't got enough sleep and this stuff makes me crazy especially when I'm tired
Sean
Honestly, it's like shooting a fish in a barrel. Twice. With an Elephant Gun. At point blank range. In the head. -
So how can their tests be valid if they don't have someone with any knowledge of the other system? You can't compare motorcycles and bicucles if you don't have someone who knows how they both work. If I put water in the gas tank of the motorcycle (since I didn't know what I was doing) the bicycle would easily outperform the motorcycle.
Don't leave your mind so open that your brain falls out. Don't close it so much that you cut off the blood.
Mindcraft says in the report that they had Microsoft's help in tuning the NT box, but that they got no help in tuning the Linux box. Others have said the Linux box was in fact 'de-tuned' from the out-of-the-box settings, which further worsened performance.
If that isn't skewing the test in NT's favor, what is? Other reviewers have done side-by-side compares with 'vanilla' installs, and in those cases Linux blows NT away. Let's write off Mindcraft as Microsoft PR, and try and get some more 'real-life' performance comparsions...