Slashdot Mirror
WSJ Says Linux Lags
TroyD sent us a link to
a
WSJ Article on Linux
that Says Linux is Good, but that it lags behind its rivals.
Troy also sent a choice quote from the article:
"...Linux currently lacks some of the features
demanded by corporations. [...] Among them are the
ability to run
simultaneously on many processors in a single computer and
to keep a log of what the computer has done."
Cool. I can save a lot of diskspace: rm -rf /var/log.
"HTTP/1.1 Server Too Busy" is what I'm getting now on www.msnbc.com
/., and then getting Slasdotted withing minutes... Duh.
www.msnbc.com Slashdotted????
Geez, which OS has shortcomings now!
Spreading FUD, getting linked on
Nuff said.
You need a better search engine. It's http://www.dhbrown.com/
What with people going ape-nutso over any and every high-tech stock these days, this story could be seen as a blatant attempt to force a short-term rise in Sun MS stock.
And it looks like it worked.
Why don't I say IBM stock? Well, there is only so much you can do, even in the hype sphere.
Phil
In a single box, it becomes inefficient to use more than 16 processors. This is where you should use clustering instead of SMP; IIRC, there is no limit on how many boxes you can add to a Beowulf beast.
Journaling isn't logging, it is a way of storing changes to a filesystem: the fs layer doesn't overwrite stuff, instead it stores the changes to the files into the empty area of the disk so that when a crash happens it is easier to recover the filesystem - fscking becomes extremely fast.
-- Jakke
They just bought part of Red Hat Dell Press Release announcing equity investment in Red Hat
If the authors actually did claim that Linux wasn't able to log correctly or that it wasn't able to run on multiprocessor systems, then they are plain wrong. End of story. It doesn't matter what kind of reputation they had or what they've done in the past, they are incorrect now and that is all that matters. It is as simple as that.
It hardly takes a great deal of research to prove both statements wrong, and if indeed they made both statements (and MSNBC wasn't quoting them out of context) then their entire report is suspect. There is no other conclusion you can reasonably come to. They simply did not do their homework.
nuff sed
They're just doing that to kiss the @rse of bill gates.
Think about it Dell have always been in bed with MS so if they invest in RedHat then it looks like that there is competition in the OS world helping MS in their trial.
Anyway why can't companies invest in someone OTHER than RedHat. There's plenty other distributions out there and some are much better than RedHat.
according to the report abstract ate DH Brown people compared Red Hat 5.2 and Caldera OpenServer 2.2
http://www.dhbrown.com/dhbrown/linux.html
th
These dont use version 2.2.x of the kernel do they?
And before 2.2.x the kernel didn't support SMP very well did it?
Are there any distributions that use 2.2.x yet?
Perhaps this isn't quite FUD
Alternatively, one could write a ufs/ffs for Linx, which shouldn't be hard given the number of open source versions of it, and then get softupdates for it.
With kickass XFS support
They probably also found in "The Study" That :)
SGI would actually be an ok system if only they
could do Graphics
JWitt
softupdates is not quite equivalent to a true journaling FS... read the disussion in the linux-kernel archives for more info.
It would be nice to have better support for BSD filesystems in Linux, though.
I blame Lee Gomes, the author of the Wall Street Journal article. He wrote the falicies, and clearly did not do his research. Who cares about the survey that was used, Gomes did not look at the context or broader picture. He seems to be the tech guy at the WSJ, writing most of the Linux and computer-related articles, and I think he lost a lot of credibility because of this article. For that, I think he deserves to be flamed :). lee.gomes@wsj.com might work, but bigfoot turned up leegomes@aol.com. Email both? I did.
The Journaled File System (JFS) was created by IBM. JFS adds a logging technique that helps recreate the file system control structures and restore the file system to a known state in the event of a crash. Other goodies in the package include the ability to dynamically extend filesystems, striping (spreads fs across multiple disk drives with alternation track patterns, allowing concurrent I/O).
To sum it up, JFS logs every operation, to a specified disk (/dev/hd8 by default on AIX). When the operations are completed they are written to memory. This is called the "sync" log. Whenever the system reboots it reads the sync log grabbing the last operation, (which in most cases is a shutdown). If there are any indications in the sync log after the last sync record, it indicates a crash, and doesn't start applications that did not have a commit record in the log, since their state may be incomplete.
Here's another gem from the unbiased source for the WSJ article:
>This paper finds that corporations derived >significant added value from Systems Management >Server (SMS), even if they were using a >framework solution.
I'm not a system engineer myself (but I play one on TV, right?) but everything I've heard about SMS is negative. All the reports I get from friends that have installed and use it on their systems is that it is very time consuming, and thus expensive. Putting a rosy face on M$ lame products seems to be one of the main line items in D.H. Brown Associates' budget. There are other reports lauding COM (M$'s attempt to kill CORBA) and at least one (Microsoft's Holistic Enterprise Strategy) which is just a puppet's rendition of a Microsoft press release.
In short: Biased? No, not a bit, just a paid promotion company for M$.
Now how hard could it possibly be to patch Apache or any other open source http server to identify itself as Microsoft-IIS?
Hmmm...kind of makes ya' wonder about how widespread the usage of Microsoft-IIS really is...
It's been discussed several times before on ./
Logical volume manager and journal file system.
If you haven't used them then you have no idea what you're missing. I hate, absolutely hate managing systems which don't have these features.
You know, it's funny... I've been playing with a lot of network intrusion detection devices oriented towards the NT side, and I can only think one thing: "All the information these tools give me is contained in a standard log on Unix."
If in fact MSNBC *WAS* Taking these statements out of context... It has the same effect as the report
being inaccurate in the first place.
What are people gonna read? The report, or the MSNBC Summary/Article? Although it does matter if FUD came from MSNBC (Or if it was intended by the original authors), the fact is it still ended up as FUD at the outcome...
JWitt
Then Linux is the big winner. Taken from the report :
Official Red Hat Linux 5.2 : $50
Caldera OpenLinux 2.2 : $50
Windows NT 4.0 Server Enterprise Edition : $3999
Conventional UNIX systens : est. $1000-$5000
And for those NT bashers out there. The review meantions compares Linux to an NT with extra expansion pack such as Microsoft Cluster Server, etc.. that make it even more expensive.
I think this article is a good opportunity to see "The Road Ahead"(tm)
p.s: read the executive review of the report by D.H. Brown, it is indeed very well informed and interesting (and it DOES mention Beowulf, SAMBA and so on)
Exactly,
I immediately went and downloaded the Executive Summary... and read through it.
DH Brown is not making the statements that the obviously technically challenged WSJ writer attempts to paint. They criticize Linux's SMP support and memory support (in relation to Solaris or HP's). They are concerned at its current lack of commericial clustering (clustered db's), high-availability, LSM, journeled file systems, and current crop of high level (PeopleSoft, SAP) application support. Exactly the things we're interested in... I see the DH Brown article as more of a roadmap for what we need to tackle next.
The WSJ article was just plain stupid... I encourage everyone to semi-politely email them and let them know that you would prefer a bit more professional acumen in its technical reporting.
Granted people are pretty stupid, but I still think that joe idiot can still put 2 and 2 together and realize that MSNBC might be giving a distorted view.
Congratulations,you are a moron...
Executive Summary available for free download.
Boy am I glad that some of you aren't responsible
for anything I care about with the attention to
detail you give.
Ok, the article refers to a study by D.H. Brown (http://www.dhbrown.com/dhbrown/linux.html).x 2.html), which shows similar results except for Linux, which was rated second best OS, behind AIX.
Strange, but a little earlier in 1998, another study was made by Datamation (http://www.datamation.com/PlugIn/newissue/02linu
I wonder why Linux sudenly "lost" so much ground, when the other OSes keep the same rank...
HP-UX usually uses VXFS, which is journalled. Sun can do journalling too if you use DiskSuite.
And, yep, the good 'ole logical volume stuff is nice. Just stick a new disk in, add it to the volume group, and extend filesystems as needed. Or move all the physical extents on one disk to another runtime and pull it out. This Would Be Nice on linux, even tho PC hardware often doesnt deal too well with the pulling or addition of disks runtime anyway.
Digital's AdvFS, HP's Vxfs (I think that's its name)... Solaris with Veritas. LSM is great, and frankly I'd like it at a workstation level as well as server. (On a server its practically critical)
A couple of your friends that run Linux at home call the OS "Slowaris",
:-).
do they? Tell you what: to find out if they have any clue whatsoever,
why don't you ask them whence that term originated. Bet they don't
know. (The fact that they've using the term is a strong indication
right off-the-bat.)
Solaris' X is slow? Where do you get *that* from? Solaris' X never
crashes? Right. Lemme tell ya: Netscape can crash *any* X
The interface is ugly? What? It doesn't have themes, does it?
Themes. Gack. Give me a nice, clean OLWM/OLVWM desktop any day. "The
interface is ugly" is the kind of thing I expect from Ms-Win lusers.
Ms-WinNT is a router by default? Are you *sure* about that? I know
that Ms-WinNTW machines aren't. And I *suspect* that Ms-WinNTS isn't,
either. (Don't remember. Been over a year since I was condemned to
installing one.) Even if true: it's inconsequential. All *nixs have
stoopid out-of-the-box configs too. You fix 'em an move on. Or you
don't and get bit.
Talk about FUD.
I've been trying to get the journaling working on my NT box for ages.
Where is the documentation ?
What the article states is clearly false. Sure you can bend the statements to make them true, but what it states in clear plain text is false. Don't give Mr. Gomes any credit; he doesn't deserve it - he didn't do his research.
Ok, the article refers to a study by D.H. Brown.
Strange, but a little earlier in 1998, another study was made by Datamation, which shows similar results except for Linux, which was rated second best OS, behind AIX.
I wonder why Linux sudenly "lost" so much ground, when the other OSes keep the same rank...
As a former DJ employee, I know what his email address should be: lee.gomes@dowjones.com.
Whoa there. IBM does NOT lead the way in SMP, Sun does. The E10K supports 64-way SMP. I don't want to get into a NUMA-Q vs SMP religous debate. IBM won't get you 64-way on ONE node, but Sun will. After we starting adding more than one node, YMMV.
Being a former DJ employee, I worked on the server end of "NewsWires" backend system. During my "tenure" there, most stories were written onto one computer system and then eventually transmitted to any one (or more) of its publications (WSJ, Barrons, etc...). The computers used to transmit these stories are running LINUX.
Since the story did come from WSJ, there is a very good chance that its transmission was made possible BECAUSE OF LINUX. So, no matter what this person says, his story (and any others that were published in WSJ) may not have been published if it weren't for LINUX.
Just a thought I had while reading this article and seeing how naive this person really is when it comes to the world of computer software systems.
Think about it. Talk up any security hole, crash bug, whatever in anything that comes "bundled" with NT (hmmm... I don't recall asking to install a flaky web server) as though it still exists and is a serious problem. What's their defense? "No, that was fixed in SP3. No, that was fixed with HotFix blah." Even if those bugs are fixed, talk about the scope and significance of them.
:)
For commercial software, having the existance of past nasty bugs exposed and talked about is far worse than any lies that could be made up. Nasty bugs in a released product reflect either poor testing or poor design and can't be dismissed.
So, forget spreading FUD about M$. Just spread the truth about thier track record of performance, bugfixes, and reliability. You give them enough rope, and they'll hang themselves.
Speaking of rope, anyone else looking forward to seeing how much rope comes as an "integrated" part of Windoze 2000?
Here's a copy of the email I sent to D.H. Brown Associates:
To: gina@dhbrown.com
Subject: Real researchers back up their conclusions with facts.
>From http://www.msnbc.com/news/256197.asp :
"But he said Linux currently lacks some of the features demanded by
corporations that intend to run their entire business on computers.
Among them are the ability to run simultaneously on many processors in a
single computer and to keep a log of what the computer has done."
Congratulations! You have just assured that my company will never
purchase any of your so-called "research". I will also strongly
recommend to all of my clients and associates that they refrain from
purchasing your brand of ill-condsidered misinformation.
I suggest you terminate your relationship with Mr. Iams, as well as his
supervisor, along with whomever it was that hired them.
Do your homework *before* you hand in your paper.
-Michael Craig Amper
Systems Designer
TekConnect Corporation
michael.amper@tekconnect.com
You people don't know anything. Instead of disputing the author on what he just said, you just make excuses for why he is wrong. Most of you don't know anything about what he said, yet you seem to be linux "experts". Well are you?
I equate you to the captain of the titanic standing on his ship and refusing to acknowledge his boat is sinking because it just can't sink
Who wants a GUI for there logs?? I'd rather stick with my awk, sed, grep, and perl thank you very much.
Linux 2.4 is supposed to have the ability to "log what the computer has done."
This is really exciting.
Whats next? the ability to compile source code to a machine readable format?
We live in an amazing world.
Quick questions... where did Stu mention NT? I can only see him referencing to AIX and Solaris...
Why does Linux have to compete against NT all the time? Why not try to take on something a lote more similar, and a lot harder of a task, like bringing it up to Solaris/AIX standards?
I have one thing to say. Melissa. Our entire NT side of the house shutdown because of one little macro virus. The unix side couldnt care less. We had to stop our mail server so the NT guys could check and clean their machines before getting infected again because their users are click happy. I have used both and find NT to be a great gaming/multimedia machine. Linux is much better in getting programming/hacking done. NT probably better in getting a spreadsheet out. All I know is we have a few OLD linux boxes running as webservers/DNS servers that have been up 140 days with out a reboot. Lets see NT do that.
an observation... if you scroll to the top of the whole WSJ article, it says "Sponsored by Microsoft" -> so the whole article is just following the M$ party line... its amazing that they let them admit as much as they did!
johnrpenner@earthlink.net
Is this your response to the WSJ report on the D.H. Brown Associates report or is it your response to the D.H. Brown Associates report? Do you know the difference?
There.
I learned something from that post.
Thanks,
-k
100Mb ethernet interconnect not fast enough for you? Try channel bonding them, or use ATM.. gotta love it.
NBC
Quite a few online papers and other news organs have taken the Brown study and used it as the basis of "Linux lags ..." stories. This is FUD.
It's not ignorance. Some editors are looking for any excuse to counter the generally pro-linux press coverages recently.
Why they are doing this, I don't know. However, don't be surprised by more of the same whenever anything which can be interpreted negatively is reported by Gartner Group, Brown, and other IT industry consultants.
Partly, it may stem from pressure from several parties concerned by the reaction of the Linux community (Slashdot, Free Software Foundation, Bruce Perens, etc) to various semi-open source initiatives by Apple, Sun, etc. These companies and others (not so much MS, I think) may be the real opponents of Linux because they hope to pull developers away from Linux and GNU projects into their own semi-open endeavours, which currently are unattractive by comparison, and may regard Linux as a competitior to their systems and services. This is especially true with Sun and Apple. These people sell operating systems that are semi-open and in direct competition with Linux.
These mainstream computer journalists always respond to pressure from advertisers. I question the wisdom of Taco publishing their articles here, but when he does they always get a good response. It is always pointless to counter the specifics of such FUD on a point by point basis
which just gives legitimacy to patently false statements and more fruitful to discuss the motivations for such articles.
Other computer news sites like Yahoo are more objective and I suppose it's a matter of what one considers to be newsworthy. What do you expect from ZDNet, C/Net and MS sites? But don't think for a minute that much of this is due to "ignorance" on the part of these ezines or editors.
don't play their game!
nuf said
The D.H.Brown report used the industry standard TPC-C and SPECWeb96 benchmarks to evaluate SMP performance. They obatained their data from www.tpc.org and www.spec.org. Apparently there was simply no published data for Linux, so it wasn't included in their benchmark comparisons.
Beowolf is irrelevant in this context
If you read the "executive summary" (16 pages) of the D.H.Brown report, which you can download from their web site, you'll see that it was actually a very fair, informative and generally clued-in report. The problem is really the WSJ article. When I e-mailed D.H.Brown to alert them to the article, their reply was that the WSJ "oversimplified a bit"; having read the summary, I think that's being kind.
The MSNBC editorial comment before the article was surely about as awful an MS-doctored spin you could put on the report. The conclusion of the report was that server OS's basically rank from best to worst as: commercial Unix, NT, Linux (read the summary before whining about it). MSNBC's spin on this was "Linux is worse than any commercial OS, including NT"... no mention of the fact that D.H.Brown found NT to be a poorer performer than even the "weakest" commercial Unix.
Maybe LEE GOMES of the WSJ is trying to get an
audience for his panel with Linus next week at
AIIM '99 -or- maybe publicists TSI who promote
the Gartner Group gang are trying to stir up a
little spice for their other client IDG LinuxWorld. Hard
data or not, it doesn't stop report selling
research companies, techno-journalists and
high-powered PR firms from going about their
business.
Most Wide and UW SCSI cards _do_ allow hot-plugging - the bus simply has to be quiescent. UnixWare enables you to halt the SCSI bus operation and change devices (add/remove CD's tapes and HD's etc). I've done this on UnixWare 7 a couple of times (this isn't properly supported in unixWare 2), although I think there are some limitations when you use Volume Manager.
Several SGI/Crays also do 64-way SMP per node.
So write it!
Jeez, this is an annoying response.. Everytime someone wishes linux had a feature someone has to tell them to simply "write it!" Yeah, like your average user who wishes for a pretty wordprocessor is going to take years to learn how to program a large project just for that.. Not everyone is a programmer, and really, not everyone should have to be one. I'm all for contributing, but if my mom wishes her scanner would work in linux i'm not going to hand her my book on device drivers and tell her to "get coding". This is off topic i know, but its a pet peeve..
Keep personal feelings out of it -- whether you agree with him or not, he has a valid point.
- Speed
It may not be a very substantial message, but did it need to be moderated?
which is the easiest way to make the linux community look like an angry hive of bees.
8^)
Mark
See the forthcoming reply below.
-michael
unless you have $995 to shell out, you can't get the details. but the derogatory headlines come free. that's no fair...
Apparently YOU didn't read the report. *I* am amazed by the stupidity and audacity of people who feel they can make snap judgement calls concerning the experience and reputation of people they know nothing about based on a single statement.
You, Mr. "Electric Eye", have completely misread what I wrote. Exactly what are your qualifications, Mr. "Electric Eye"? Why do you hide behind a link to judas-priest.com? Do you actually make a living off of selling your expertise and reputation in the technology fields? Are you afraid we might find out who you really are?
I don't know about you, but I think the work that I perform designing networking and systems solutions for "multi-site, multi-departmental (to use DHBA's definition)" enterprises qualifies me to make judgements concerning the quality and veracity of this report.
I admit that my quotation of the MSNBC/WSJ article was rather rash, but this still does not excuse the poor quality of D.H. Brown Associates' supposed "research". I over-reacted to the article because I am sick of the media coverage given to these psuedo-research institutions whose main goal seems to be to attempt to downplay the significance of any new technological phenomenon in favor of the status quo.
I make my living working with leading-edge technologies, and it seems like every other week I'm fighting off ridiculous assumptions like those posed by the DHB report. Misinformation of this kind strikes directly at my bottom line, and I'm tired of it.
I read the excutive summary just like most of the rest of you, and found the methodology of their supposed testing sorely lacking. Many of the claims that the paper makes are based upon the availability of "published" results rather than true scientific testing. This is what I meant by stating that "Real researchers back up their conclusions with facts". Nowhere in the summary does it refer to any actual testing performed by DHBA.
It is a fact that D.H. Brown published a report that lacks a factual foundation for some of its conclusions. It is also a fact that I can back this up by referring you to the original document. I certainly will not claim to have read the full report, since after reading the summary, one would be very foolish to spend a thousand dollars on the full report.
If YOU had read even so much as the summary, Mr. "Electric Eye", you might have noticed such gems of wisdom as:
"...as the purpose of this study is to gauge the suitability of enterprise deployment, such products must be fully supported by a credible supplier."
and
"Since such tests are not a part of this study, stability does not warrant a formal rating in this assessment."
It seems to me that the statements of this type that pepper the summary (and there are more than these two) would seem to render the entire report irrelevent.
What I find most interesting is the glib assessment that Linux is not as capable as Windows NT or the commercial variants of UNIX without a clear definition of exactly what is meant by *enterprise-level* computing.
Most *enterprises* would be hard pressed to find a data file larger than 2GB or a need for more than 2GB of RAM in a single machine. Should Linux be considered less capable for *enterprise* computing simply because it fails to meet the requirements of a very few organizations who deal with extreme volumes of data?
Perhaps Mr. Iams and his associates should look up the Department of Commerce and ask them for statistics on the commercial value of large *enterprises* versus the rest of us.
Should anyone else feel it necessary to comment, kindly do us all a favor and keep the personal attacks to yourselves. If you aren't talking about facts, you're just talking shit.
-michael.amper@tekconnect.com
Anyone have any opinion of SFS on the amiga ?
Would it be at all useful for linux stuff - most of the problems it addresses are amiga-specific, but some of it is interesting from a linux perspective too. It seems to be a journalling FS which uses extents, and never invalidates as such ( the bane of amiga FFS users - l:disk-validator kicks in on bootup, a bit like fsck, and there's no stopping it...) They're working on merging the amiga mufs v2 spec( multi-user filesystem) into it too...
http://www.xs4all.nl/~hjohn/SFS/
Journalled filesystem for Linux is being worked on. I havn't got a link (Someone..?), but it was
discussed here a while ago in a "Ask Slashdot" feature about large discs. It really *is* a cool
thing by the way.
Look at: "Among them are the ability to run simultaneously on many processors in a single computer"
So the question is : Does Linux really run _simultaneously_ on many processors ??? When he refer to Linux it must be the OS itself.
Does the OS really need to run on multiple cpu's ?, NT probably does since it eats cpu cycles whenever needed.
Just a quick though...
I wonder if people are confusing running SMP benchmarks well with running actual applications.
Multi processor appears to be bandied about a lot by people who don't really understand the issue
doesn't disksuite still require you to manage the
disk slices and metadata by hand
under AIX all that stuff is transparent
There are issues with Linux, like shipping security out of the US that commercial OS's can get around with licences from the govt. That's a big problem for Linux - you can't just download, or buy for 2 bucks, a SSL enabled Web and News server. You can't even get it for $3000. Not that it's hard to setup mind - I've done it myself - configuring Apache for SSLeay was quite easy, but that's not what DH Brown measures - and it's not something that can easily be measured (unfortunately, for the free s/w croud).
ApacheSSL is widely available outside the US. If I download it from Germany and I'm in the UK then I'm not exporting it from the US and therefore I'm not breaking any US laws. The same applies to ssh, pgp, ...
Given you are staring for 3 minuters at "welcome" I reckon you must be using something like a sparc2... How long does X take to start up on a 486-30, then? /bin/bash*
If it really is the appearance of the window manager that's bothering you -- why not try to build something nicer on solaris? A graduate student at our place just built KDE - which, admittedly isn't exactly for the beauty contest either, but I think gives an indication of what's possible.
I also cannot quite follow your problem with executables - even if they might be somewhat larger for risc architectures:
-rwxr-xr-x 1 bin bin 396188 Oct 21 1997
This would be the same, BTW, with linux-sparc/-alpha/-ppc...
Finally, with what would you maintain your cluster of PCs you could buy for $12000 a year? Hate to bring up that "total cost of ownership" thing, but even Linux PCs don't maintain themselves alone - and from my experience, sparcs seem to be less demanding of manpower - which is, after all, usually the largest expense factor.
http://www.dhbrown.com/dhbrown/linux.html
D.H. Brown has ZERO credibility. This is just dreck written by salespeople.
I think the problem with this article is that, based on the introduction, and the tone of the article in general, the statement about Linux not scaling to many processors seems to have an implied ".. when compared to Windows NT" attached to it. I don't know if this is what the author indended, but in any case this is not true.
Linux scales better than NT, at least for 2-4 processors (my limited experience). Of course there are commercial Unixes that scale to hundreds of processors, which linux doesn't do - yet - because it isn't a big issue for the vast majority of linux applications - yet. If you're going to run a 50000 user database server, you won't run Linux - but you definitely won't run NT either - the choice will be AIX or a Sun Enterprise server or something like that.
There is some info here and here if anyone's curious.
Nah... An infinite number of monkies pounding on typewriters could turn out Shakespeare.
Microsoft? Sixteen monkies sharing a fountain pen and a case of beer.
Take away the pen and you've got ZDNet.
Replace the beer with Jolt and you've got Slashdot. ;)
-D
dcross@cryogen.com
...is the author of the article.
:-), but judging from the article I could read, this guy does not have a clue. And thy shall not write about things you do not know. I would love to email him and explain some of Linux features and I'm sure that some of you would fancy that aswell, but I cannot dig up his email adress. Please help! Dig dig dig! and post his email as a headline! A password to the articles would also be nice. /Patrix
I ran a search on him at the archive on wsj.com and came up with these headlines:
"Linux System Performs But Has Some Problems"
"Linux System Makes Waves But Has Limits"
"Linux Operating System Gets Big Boost From Support of H-P, Silicon Graphics"
I could not read any of them since you need a login (did try cypherpunk of course
The study compared NT and Linux. So they choose the playing field. Neither system is even slightly prepared to serve at the very peak of enterprise level work.
The WSJ article claimed to be looking for systems suitable for "the worlds toughest jobs". It should have disqualified both Linux and NT. Since it did not, we can dismiss their assesment as flawed at best. To go on and suggest that NT was "better" than Linux by using a study that did not purport to study the stated assumption, suggests FUD.
From what I hear, the WSJ is a Microsoft first shop. If they can't beat a developer into using NT, then they'll settle for Sun. Using Linux is evil and career limiting. Soon after said career is limited, it will be ripped out and replaced.
The underlying study is actually a very responsible view of how Linux stacks up against the highest-end commercial OSs. It says that the main shortcomings of Linux are that it lacks proven SMP ability (we're talking 6 or 8 processors here), file system journaling (which is what the clueless WSJ meant when it talked about "keeping a log of what the system has done"), and very-large file support (like, on the TB scale). The big problem with SMP, apparently, is that it takes years of working hand-in-glove with the hardware mfgr to get a properly tuned SMP system. Probably correct.
While the study ranked NT at about the same level as Linux, it expressly did not consider stability. The study said that stability evidence was almost entirely anecdotal, and there were no real MTBF studies to review.
The version I read in the Globe and Mail was pretty terrible: it had a worrisome headline, ("Linux not quite up to snuff") six generally accurate and positive paragraphs, but its author obviously ran out of brains before column-inches, misquoting the study as has already been demonstrated.
I think this was a case of "balance": if something's good, we have to say something bad about it. Unfortunately, they said the wrong things, because they didn't understand what the study was telling them.
The Wall Street Journal ("WSJ") is a "journal" (diary) about Wall Street. It is very popular with people who trade stocks and invest -- indeed, it is more popular in some circles than the National Enquirer. Unfortunately, WSJ isn't printed in English, and doesn't offer home delivery.
Mind the Gap
If the person were transliterating, we could try to sound it out and perhaps determine what was actually said. Unfortunately, what we've got here is an incompetent translation.
It's kind of like calling "Halt, stranger, or die!" a greeting. It's not entirely wrong to do so, but there's a great deal of important detail that's been ignored.
Mind the Gap
My biggest beef with NT is the filesystem. How can you have a system where when one file is corrupted, say NTFS.SYS or other important file, with a filesystem that you cannot write to from DOS or anywhere, you have to reinstall the entire OS (the repair bit fails many times)?
It's ridiculous how anyone is willing to put their entire company on that OS...
"NO ONE paid for this study."
Heh, it would be funny if no one bought the full study as well. All that work, and no $995 checks to make up for it.
Anyway, no one paid for the study beforehand, but they are in it for the money, obviously.
The article makes some interesting assertations and provides some good constructive criticism, but unfortunately the criticism was dumbed-down to the point of creating misinformation about what Linux CAN do properly.
They mentioned logging capabilities, and to that I assume that they are referring to a journaling file system, or audit trails, both of which truly are missing on Linux. The trouble with the assertation was that it was too broad, and I believe this stems from an attempt to write about a technical subject to a non-technical audiance. The average WSJ reader doesn't know or care what an audit trail is, or what a journaling filesystem is and why you'd want one. The author's solution was to simplify to the point of PHB understandability. Unfortunately this led to an article that gave the impression that Linux has NO logging abilities, something that would petrify any reasonably responsible executive.
The SMP allegation is a bit trickier for me to understand. I would presume the author was referring to big-iron 64 and 128 processor type scalability, however this article also implies that it lacks even 2 or 4 way scalability, and implies that NT has greater scalability.
I'm not sure how best to respond, I wrote a paper letter to the WSJ since I'm a subscriber to the print edition outlining my concerns with the article and the possibility that there was some misunderstandings due to the technical nature of the article. Hopefully things like this will eventually disappera.
The most important thing is to keep focus. Write code. If you can't write code, test code. If you can't test code, write documentation. Just fine something, even the smalles thing, and do it. Anything we do to work on improving Linux and the related applications will solve many more problems than explaining to reporters the errors of their ways.
Yeah, Linux will do per-processor accounting. Try "man acct" and see what it says. The author of the WSJ article and even you can't seem to RTFM.
The fact that Linux and the free BSDs are smashing Windows NT's house of cards at every turn is enough to shoot down this little FUDlet quote.
To say Linux equals even the best commercial software is more of an insult.. After having used Linux and other assorted free OSes for 4+ years, I'd say they surpasses it by a longshot.
I'm talking about within the confines of x86 hardware. Linux and FreeBSD are by far the best operating environments available for x86 in my opinion.
Well, with SGI/Cray out front. 32 processors? Try closer to 265 and beyond. NT doesn't scale well beyond 4 processors, and even at 4 you can see the scaling issues in NT. Worse, the Intel x86 arcitecture leaves a lot to be desired, especially when you start scaling to tens of processors. Just look at the kind of contention you see on a PCI bus on a single processor system running a moderatly memory and graphics intensive process. Even if you scale a Linux/x86 box to XX processors, you may not be able to use most of that power because the chips are always waiting for memory/IO/communication with the video or audio subsystem.
I read the internet for the articles.
As to logging: Most free Unixes do a pretty good job of logging, although most commercial Unixes do better. The difference isn't substantial, but it is the kind of stuff you can only do with intimate knowledge of the hardware. As compared to NT, every Unix flavor I've seen blows the pants of NT w.r.t. logging.
AFAIK Linux doesn't have a general system level statistics logging ala Performance CoPilot. This makes Linux a little more difficult to determine exactly where the bottleneck is on a loaded system. (Is my system slow because I'm out of System Time, or am I always waiting for disk I/O, or maybe the PCI bus is jammed up?)
As to your aside: most big systems spend a very small fraction of their time with all of the resource management tasks, and these tasks are necessary when you have to determine what to add to a system. It is extremely difficult to determine where the bottleneck is in a system when you don't have statistical information on every subsystem over a period of time, especially when the bottleneck isn't something obvious like memory or CPU time, but something more fundimental like contention on your SCSI busses or memory bandwidth issues.
I read the internet for the articles.
According to Netcraft:
www.msnbc.com is running Microsoft-IIS/4.0 on NT3 or Windows 95
Well there you have it. The silly website is running IIS v4.0!
NJV
Posted by LordPraetor:
I would certainly run a 50,000 user database server on Linux... I just wouldn't use the standard login apps. Depending on what you are doing, my Dell PowerEdge 2300 running v2.2.3 and MySQL could handle a db of that size easily. I wish I could say the same of my Dell PowerEdge 2300 running NT4...
Posted by LordPraetor:
Like a graphical interface? (X)
Nuff said
Posted by !ErrorBookmarkNotDefined:
This letter goes directly to a source used in the article. This letter is directly on point. It clarifies one of the central questions about the MSNBC article: "Where did this information come from?"
And it gets a score of 2, just a hair above some AC posts?
-----------------------------
Computers are useless. They can only give answers.
Posted by patg:
Has anyone refuted them? Is there any way to publicly show them to be in error?
This falls into the disinformation category. This is about the most uninformed article yet.
Posted by Nino the Mind Boggler:
"This article is automatically bad just because it's posted on MSNBC."
C'mon folks. We have to do better than that. I don't know the details on what Linux is and is not capable of, so I need some assistance sorting out the FUD from the valid criticism. What is Linux capable of as far as SMP and logging? What are Linus, AC, et al working on in those areas?
I noticed Alan Cox mentioning on the Kernel mailing list that Linux 2.3 was going to add many of the features mentioned in the above post, including support for more then 65k users. Sorry I dont have the exact URL.
Someone needs to send a very well reasoned letter to the editor of the WSJ, with a cc: to half a dozen other (reputable) sites including the original (FUDdy) WSJ article and a point-by-point refutation (including sources). If WSJ doesn't post the letter, then the other half a dozen sites have a story to scoop: "WSJ screws up facts, fails to post rebuttal."
Who am I?
Why am here?
Where is the chocolate?
What is your Slash Rating?
MSNBC have run quite a lot of articles in frank praise of Linux and critical of Microsoft; they're nothing like Slate who really do seem to be an openly biased FUD-organ.
This article isn't theirs; it's a reprint from the Wall Street Journal, who are in turn quoting a report by a "consulting group" that no-one's ever heard of and that clearly don't know a damn thing.
--
Xenu loves you!
The other, and often overlooked issue, is Linux's ability to run on >> 2-4 procs. For many larger businesses, their core systems run on many processor machines that are designed to be extremely fault tolerant and stable (up times > 10 or 20 years are common). Beowulf-style clusters aren't really an option. The article (which was posted yesterday) is more about the appropriateness of Linux as a replacement for S/390 and AS/400 type servers.
>The first problem: "Run simultanously on many processors" has been answered by SMP.
According to the SMP FAQ, Linux's SMP support has been tested on 4 processor systems and theoretically should support up to 16 processors. While this is a reasonable number, it probably falls short of being "many."
Ooh, a sarcasm detector. Oh, that's a real useful invention.
I agree! This article should be the top-scoring article in this thread!
Computers are useless. They can only give you answers. -- Pablo Picasso
I'm reading the exec summary of the report right now, and I have a question for anyone out there w/ experience in various unixen and nt.
First, the premise of the report is that Linux is not enterprise-scale material, right? Okay, the intro paragraphs from the summary end with this conclusion - after mentioning the oses.
And I quote:
They(the two Linux distros) both fall short of the coventional production-grade implementations of proven, non-trivial SMP scalability; high-availability clustering capabilities; journaling file systems; logical volume managers; large files; and many other less significant but useful functions. end quote.
Uh, and NT Server 4.0 Enterprise Ed. has all this? Can someone tell me if this is true, yes/no?
Another question. This summary doesn't seem to make any real judgements about the oses in terms of how well they work, what you need to get all these wonderful goodies in terms of hardware resources, training, administration. I find this rather a large oversight. I mean, MS might be able to offer everything, but how well does it hold up, and what kind of resources do you have to throw at it to get to this scale; what about TCO, ROI, etc.? I agree that enterprise scale needs big iron, and Linux ain't there yet, but I also believe that nt ain't there neither. I think I have a problem with this approach. Also, the clueless journalists who "summarized" the report, or summarized the summary most likely, really should have kept their fingers off the keyboards and on their dicks and just provided a link to the free pdf file.
"shop smart:shop s-mart" ash
A report by D.H. Brown Associates is used as a source for John Kirch's NT Server 4.0 vs. Unix page. They don't seem to be too biased. On the other hand, I couldn't figure out how Linux came in under NT. Even if they were talking about journaling, NT doesn't do that either, and Linux DOES do SMP. They either got their facts screwed up, or they had some reason that they didn't present in the article.
It's not enough to bash in heads, you've got to bash in minds. - Captain Hammer
As near as I can tell from the executive summary, Linux lost to NT mainly because there wasn't enough information available about its performance. The requisite studies and benchmarks didn't exist for Linux, so it lost in those categories by default. The executive summary claimed that NT has a journaling FS. Is this even true? The only other thing that NT seemed to win out on is clustering for web and database serving.
They apparently went strictly on functionality rather than price too. While they do tell you that Linux is can be used to build a Beowulf supercomputer, they say that NT can do basically the same thing using public domain software packages. What they forgot to tell you is that for the price of all the NT licenses you'll need for that, you could buy a real supercomputer.
I think the report tried to be objective, but I also think they cut NT a lot more slack than they did Linux.
It's not enough to bash in heads, you've got to bash in minds. - Captain Hammer
They used OpenLinux 2.2 which uses the 2.2.5 kernel for the review, as well as RedHat 5.2 with the 2.0.36 kernel.
It's not enough to bash in heads, you've got to bash in minds. - Captain Hammer
I believe the report actually said that Linux didn't do SMP well, and that it didn't have a journaling FS. Which, oddly enough, it said that NT did have.
It's not enough to bash in heads, you've got to bash in minds. - Captain Hammer
Didn't look at the date, but the report does claim that they used Open Linux 2.2 with the 2.2.5 kernel. Perhaps they are confused?
It's not enough to bash in heads, you've got to bash in minds. - Captain Hammer
Well, at least in the amount of space that is required and the number of crashes per day.
Misfit
How much was the Microsoft employee paid for this and how long will it take for the person who approved this article to get used to their new job title?
That's not what they said. They *SAID* linux does not do logging,
and everybody knows *EXACTLY* what was meant by
it.
They *MADE SURE* everybody knew what they meant
by it.
Read the next few lines of the WSJ article where
they tried to explain their comments...
Yeah, numbnuts, it's a subscriber-only document. The available one doesn't have a discussion of Linux.
--
Ben Kosse
Remember Ed Curry!
And it tells just about zero regarding their methodologies.
Yeah, there's more to a study than the results. You must manage something. Pity.
--
Ben Kosse
Remember Ed Curry!
Linux is a good OS. Probably the best available for the Intel archiecture.
However, it does not come close to offerings from Sun or IBM for SMP scalability, security, and high-availability [i.e. fail-over clusters].
Wanna fix this? Join the Linux-High Availability project. Write a journaled filesystem. etc. Because for NOW, this isn't FUD, there are the hard facts.
-Stu
go to www.infoworld.com and read their take on the D.H. Brown study. They focused on commercial unicies. NT was included for balance.
-Stu
I'm very glad to hear this. I think this was my point.. we need to counteract FUD by "just doing it" - creating the features the market thinks it needs.
Of course, this is for Linux 2.3... and it will be a long... long.. time before it is "produciton quality" and released in Linux 2.4/3.0 (12-18 months?) So my observations [about Linux not competing with high-end unicies] were very valid for the time being.
-Stu
- You can't run a 64-processor SMP box on Linux.
- You can't get a government B1 security rating on Linux (You can on "Trusted Solaris" or on AIX)
- Inclusive with the above, we need a journaled filesystem
- You can't get highly-available failover clusters with Linux. [though the linux-HA project is working on it]
- You can't get single-system-image clusters for scalability with Linux (Beowulf uses a low-level messaging API that essentially ties your app to Linux)
- You can't have terabyte files for large databases [that means no data warehouses]
- You can't have > 100,000 users on a Linux box for very large networks [Solaris & AIX can]
etc.
By now you're all probably hopping mad at me, but please folks: take a deep breath. Is this really FUD? Or is it merely pointing out some small nitpicks? My, my... people are so quick to criticize and yet so hyper-sensitive to their own medicine.
Let's get real: we're only talking some minimal feature-lack, and not very "widely used" features at that. Wanna fix it? Contribute code. This is how Linux makes FUD irrelevant - not through whining about the WSJ's misleading prose.
The underlying study by D.H. Brown is rooted in fact, and it means one thing: we now have specific target areas that Linux "could" be improved, provided someone with the time+need will contribute.
-Stu
They may be talking about a journalling filesystem.
It's obvious that the article is the work of a technological nincompoop who is transliterating information obtained from someone else.
Yes - NT can do this, but the system log fills up awfully quickly on an NT server if you do it...
To enable it, enter User Manager (no - really - stupid eh?), and change the Policies->Audit options.
Matt. Want XML + Apache + Stylesheets? Get AxKit.
First, here's the original "Overview" of the report (the full report is stupidly expensive). Basically what they do (DH Brown) is compare on a feature for feature basis on shipping systems. If we go on this basis, RH5.2 lacks many commercial OS features, such as Journalling, high end SMP, Transaction services, Corba/COM integration, etc that these more expensive OS's offer. The msnbc overview glossed over the report quite a bit - the report actually stated that Linux is good for a lot of things, such as web services, even for high end systems provided you have a very close fit - that's what Linux is good for.
There are issues with Linux, like shipping security out of the US that commercial OS's can get around with licences from the govt. That's a big problem for Linux - you can't just download, or buy for 2 bucks, a SSL enabled Web and News server. You can't even get it for $3000. Not that it's hard to setup mind - I've done it myself - configuring Apache for SSLeay was quite easy, but that's not what DH Brown measures - and it's not something that can easily be measured (unfortunately, for the free s/w croud).
There are some serious shortcomings in the report though. Such as looking at 2.0, not 2.2 (hence why Linux appears to fall down in comparisons of SMP, large file support, Max memory support - 2GB in 2.0). The section on SMP testing simply has a big blank space for the performance of Linux - which makes it look like it comes last at first sight. I wouldn't mind betting it's better than NT with a 2.2 kernel.
Unfortunately there's also the issue that the report just discusses the features that go into NT (and the others) that provide high reliability, such as HA clustering (which is shite on NT), resource management (also shite on NT) etc. They don't actually take into account how stable the system is in every day use. If they did, NT would come last.
I'm not sure about how Linux is worse off than commercial Unixes vis-a-vis internet services. Can someone clear up what AIX and Tru64 offer over Linux in terms of IP protocols/tools, TCP/IP extensions, bundled web browsers/servers, bundled email servers and e-commerce tools. Perhaps it's just that very last one, which comes down to issues about SSL again... Sigh.
Other than those things, I'm not quite sure how they have Linux so far down the scale. I'm inclined to believe they just got it plain wrong. What am I missing?
Matt.
Matt. Want XML + Apache + Stylesheets? Get AxKit.
I can't help but wonder how much they paid for that add dressed up to look like content...
Gumpy,
Unless Greg had agreed to let you publish his email address, it was most irresponsible of you to publish it in a public forum. Now his Inbox could get flooded with flames.
The correct way would have been to ask people to email you, and then you should have forwarded the relevant emails.
If, of course, Greg, had indeed asked you to publish his email address, please accept my apologies in advance.
There is no such thing as luck. Luck is nothing but an absence of bad luck.
Does NT actually do this? I think this is the crux of the issue. Linux can take critisism that other systems have journaling file systems. We don't like that we label as inferior to NT. Does NT logging actually allow you to see who edited files, who used programs, etc? I've never seen it under NT, but I'll admit I got my machine at working working reasonably well and tend to stay away from it's guts.
Kashani
- Why is the ninja... so deadly?
is bad
Although the NT thing is still a mystery (didn't they take into account memory useage or stability at all?)
Here's how you can tell the guy is lying. Has anyone ever seen Linus Torvalds anxious? And Linus's command of the English ideom exceeds that of most native speakers, so he would never say "anxious" when he meant "excited". No way did he talk to Linus. He's just making it up as he goes along.
-russ
Those are all valid. But so are these:
- You can't run a 64-processor SMP box on NT.
- You can't get a government B1 security rating on NT.
- You can't get high-availability failover on NT (although they're working on it)
- You can't get single-system-image clusters on NT. Heck, you can't get *any* clusters on NT.
- You can't have > 100,000 users on an NT box for very large networks.
Had they said, "Linux is good, but is still lacking features and lags behind Solaris and Tru64 Unix", I'm sure we all would have nodded and agreed. But NT??
Personally, I'm getting sick and tired of hearing people talk about Linux as inappropriate for an enterprise, and then talking about NT as an "enterprise-level OS". Sure, I'm all for criticizing Linux where it falls short. But let's have a little objectivity, OK?
(Note to Stu: No, I'm not flaming you. You're right, of course. But this "NT" thing really has me burned.)
Pure and simple. But the problem is, it's really going to affect Linux's perception in a lot of many very influential minds. Does the gray lady have an agenda here? I remember when Steve Jobs was starting up NeXT, and the WSJ bashed the endeavour really badly. It essentially shot down NeXT's enterprise market for a long, long time, until he dropped the hardware. Some of the facts here could be contradicted by basic fact-checking of the kind normally done by the editorial team in a paper like the Journal - that is, the author writes the article, submits it, and the editors will do fact-checking. It looks like this article wasn't fact-checked. I wonder why. -p
--
The real Paul Vallee is slashdot userid 2192, and, what do you mean it's not cool to point out your low userid?
NTFS does do some journaling, but I believe it's of a limited level. It is not the same as Veritas or JFS, etc.
For instance, if an NT(well or a standard Unix) partition of say 500 Gigs or larger were to go down hard. When the machine came back up, to run a full fsck on it would take several hours... You'd be down for the better part of a day or two.
But with Veritas, it can recover quite quickly without having to do a full integrity check on the filesystem, and your back up and running in only a handful of minutes.
As far as your other comment...
NTFS also allows you to extend a logical partition with additional space from another physical partition.
NTFS has some journaling, but I think it may only be of the metadata. Windows 2000 is going to include functionality from Veritas, and may result in a full journaling filesystem. I'm not clear on that part.
When the article discusses SMP, they're talking about larger scale than say 4 processors. More like 32 or so. Obviously IBM leads the way with this, with Sun close behind.
When they're talking about logging, they are most likely talking about Audit logs.
Who opened a file, who ran a program, who wrote to a file, when this occured, etc.
This is a pretty critical piece for many businesses in terms of security policy, etc.
Having a journaling filesystem available such as Veritas is also important, which is what some others assumed was being talked about.
I'm rather amazed at the number of comments calling this FUD when nobody seems to be quite sure what is being talked about. Of course this lack of understanding appears to have been encouraged by the initial poster.
Linux does run on 2 or 4 processors. That's basically the limit based on presently available motherboard technology. What Linux doesn't do is scale like say AIX or IRIX does, where you can run on dozens, hundreds or thousands of processors. I think this will be partially addressed in the coming months as the big iron vendors start adopting Linux as part of their road map. They will have to modify the kernel to work on a switch based architecture rather than a bus based architecture, which will take some time.
Again this is partially true. The contents of
I think if the Linux community decides they want to address these concerns then it will take much less time. Linux moves fast on things the community wants.
I would have to think this is true, though it didn't have anything to do with the scalability or logging concerns. NT has application support which at the moment Linux lacks. I think the ranking conclusion was misleadingly tied in with the above statements. It may or may not have been done on purpose.
Anyway, somebody should post a response (a well written response, not "you're all morons and will be the first up against the wall when the revolution comes!") that does a fair disection of the article.
many = four.
Just use 'fight training rules'. When one person encounters enemies, you count like this: one, two, lots, many, run away!
Therefore, many=four.
(ok.. so it's silly, but this discussion needed some humor).
-- There is no sig line, only Zuul.
If it's not spectacular, it's not woth writing about. So either "Linux on its way to world domination" or "Linux greatest hoax of the century" is their way to go.
I wouldn't trust anybody who has spent the last 5 years with Windows, tried to install Linux once, failed at the first attempt and now tries to tell everyone his truth.
I remember having quite a hard time when I installed NT for the first time (years ago). "Easy", "user friendly", "intuitive", they are all subjective impressions.
The parallels between the IT industry and international politics become more striking by the day.
--
As long as each individual is facing the TV tube alone, formal freedom poses no threat to privilege.
"Reactionaries must be deprived of the right to voice their opinions; only the people have that right." - Mao
No.
LILO boot: linux init=/usr/bin/emacs
In November '98 when the report came out (AFAICT)?
Wow.
They're wasting their time, they should be in fortunetelling or stocks...
a) Linux has lovely logs. Read 'em daily. Much better than the tripe you read on MSNBC. Lots of fun tools around too for gathering stats from your logs. Actually, many of the logs are maintained by separate programs. We should really say that _linux_ has a good log for kernel activity, and _wuftpd_ has a good log, and _apache_ has a good log. But all the data is there about everything your processes do. Most importantly, about connections and transfers via your services.
b) Linux runs beautifully SMP (2.1.x and above). I worked on a dual system in school last year, and my friend has one. It's very clean. And it's amusing to watch linux switch tasks back and forth between processors when it's not doing anything, looking for better load balancing. Alan Cox says Linux runs equally beautifully on up to 16 or 32 processors, and I trust him.
c) Linux does _not_ have a journalling FS that tracks everything that happens on the disk, guaranteeing your recovery of lost data, say after a hardware crash. NT doesn't have one either. Some big commercial UNICES do (AIX, what else?). I beleive Linus said it's a priority for linux 2.3.x, so maybe _maybe_ initial implementation by the end of this year.
No, you fight fire with water. And you fight FUD with the truth.
It may not always be as detailed as what you describe of Digital UniX, but as a matter of fact, one thing I frequently use Linux for is the 'hardware info goldmine' that /proc is. Combined with boot time and kernel module info, you can often find out a lot of nasty things.
It does of course depend on the driver a bit. But when a win95, nt or novell box gives you hardware suspicions, a linux rescue disk can make your life much easier!
In NT4 if you right click on any file or directory and select properties then the security tab and the auditing button you get an auditing dialog box. It appears that you can specify any user(s) or group(s) of users and any action like read, write, execute, delete, etc. Presumably this gives you an extremely high level of surveilance over who's doing what to which files.
A year or two ago I wanted to monitor access to a file on MP-RAS (NCR Unix). I needed to know who was reading the file and when. I was told by everyone I asked that it couldn't be done on Unix but it would be easy on a mainframe. (That advice didn't help much since I wasn't on a mainframe)
So, if I wanted to put surveillance on any old file on my Linux box to get a log of who read that file, could I do it? If someone just reads a file without modifying it (using more, for example) does that get logged somewhere?
Journaling is frequently provided by high-end database programs, which grab an entire partition at a time and write to it however they see fit.
Providing it at the filesystem level helps ensure reliability of all data, not just data stored in high-end databases.
I recall someone was working on a journaling filesystem for Linux - a French graduate student, if memory serves - but I can't find the webpage anymore. Last I checked was a few months ago and it was still pre-beta.
Jamie McCarthy
Jamie McCarthy
jamie.mccarthy.vg
Yes, the article is wrong. But so what? I like that fact that mis-information exists about Linux, especially in a widely read newsrag. These kinds of articles will just help create a divide between the ignorant and those that can think for themselves. It won't change the facts at all. Most corporations will never help improve Linux - they'll just jump on whatever bandwagon the WSJ tells them to. Very few corporations have a reason to help improve Linux and free software. Who **NEEDS** them? I'd rather see them spend money on NT and have a good laugh then anxiously await mass acceptance by corporations.
Of course, he was not using kernel 2.2. If he wasn't going to judge linux based on the latest kernels, why review it at all? No one uses linux "out of the box" if their needs would be better met by compiling a custom latest kernel. It's a whole different mentality from the commercial world, and it didn't seem like he understands this.
That said, who in the fuck would pay $1000 for this nonsense? If they weren't willing to even upgrade kernels, one wonders how much work they went into testing ... it smacks of "we're too lazy to do any testing, so we'll compile a bunch of info you already know, and charge you $1000 for it". What a joke. I need to start my own fudly Gartner Group clone.
support gun control: take guns from cops
Actually, it lists Beowulf (not by name, but by description) as one of the "four areas" in which Linux excells. I think that when the author refers to multiple processors in the same computer, he's talking SMP.
Christopher A. Bohn
cb
Oooh! What does this button do!?
Why don't you read the actual report before forming any judgements. I know for a fact that the authors of the study have been working VERY hard on this report. There is a *very* long track record of OS comparisons by Tony Iams that most of you will ignore, but are, and have been, taken VERY seriously by most of the top executives and developers at MS, Sun, IBM, and Compaq. Trust me. There is NO bias. Yes, this particular article appeared on MSNBC, but it is a link to the WSJ story. Claiming any sort of FUD or bias when most of you haven't even bothered to read the summary or full report (btw, NT did NOT place "first" in this study, so get off that point), is pure ignorance on your part.
The author of the report is a long time programmer (from what I know, he was programming WAY before any of you and holds a degree from Carnegie Mellon) and probably knows more about any OS than most of you guys. So, yes, if he is the one on that panel, he will be speaking the same language.
http://www.dhbrown.com/dhbrown/linux.html
Now, instead of making the FUD statements, etc. why don't you take a look for yourself. Several things, from waht I can tell, were taken way out of context in the WSJ article. Go take a look and then talk.
As I said, read the report before making judgements on it based on an article written by the WSJ. Do you really think they (the WSJ) know about Linux in the deepest technical sense? Ha!
As someone who works with them, I can tell you for a fact that NO ONE paid for this study. They've been doing them for years. That's one of the reasons you have to PAY for it.
Don't start slandering a company you know nothing about. It can get you in trouble. Why don't you do a search for a report the author did comparing AS/400 to NT a couple of years back and you will see there is no funding done by M$.
Well, genius, that IS the point of running a business. Is it not? And that's what they do. Stop trying to make it something it's not.
Apparently he didn't. I am amazed at the stupidity by some people basing their ignorant flames on a WSJ article that completely misquoted and/or misunderstood the report. Obviously, this genius shouldn't be IN the position to dictate what services his company purchases since he has no clue what the report says. Gee, that's smart. Base your purchasing decisions on a newspaper article. I doubt DH Brown would want your business anyway....
(Before I go on... Yes, I actually do make a living in the IT industry
selling my "expertise" to industry vendors. I keep my identity private for
a reason).
Actually, I *DID* read the report, hot shot. And I feel given their
methodology, the report was fair and valid. No, they didn't do in-house
testing. Why should they? What is to be gained by that? (They also probably
cannot afford to buy an AlphaServer, Sun Enterprise server, SGI Origin box,
and an RS/6000, but that's another issue...) PC Week and Info World labs do
it for everyone to read. I don't believe you will find any kind of "score
card summary" from anyone else. And considering the work Mr. Iams has done
in the past (which I doubt you have any knowledge of) I consider this to be
an excellent feature-for-feature comparison. You can argue till your blue
in the face about whether or not you agree. But I simply believe that this
is a credible study and report. What it appears to me is that, like many
others on here, you can't accept the fact that Linux isn't the
be-all-end-all fo OSes. I certainly do not discount it's power. But let's
face the facts that it still does NOT compare to AIX and many other Unices
that have been developed over the past 30-whatever years. I certainly
expect it to be an "enterprise-class" OS in the forseable future, but you
sure as hell don't see anyone running a Linux box to handle, say, an HMO's
data center. Do you? I'd call that "enterprise class."
Now, you criticize DHBA for using industry standard (read: industry
*recognized* benchmarks that have been developed over a long period of
time...) benchmarks to gauge performance. Well, what do you expect them to
do? Spend money and resources they most likely do not have to come up with
their won performance benchmarks? I'd like to see _you_ do that. Even if
they did, you'd jump on them because they were using something that no one
else in the industry recognizes as a standard! (FYI, it usually takes about
3 years of intense debate and testing by a consortium of vendors and
programmers to certify a benchmark such a SPECweb96).
Also, I don't see why you think this firm is putting down Linux and trying
to defend some sort of status quo. The study simply points out what the OS
currently lacks and where development efforts need to be focused. They
point out the plusses AND minuses of EACH OS. As someone else stated, this
study can serve as a form of roadmap for Linux.
Anyway, I am done. Enough time wasted on this topic...
Solaris is designed to run on up to 64 processors (an E10000 Starfire), each of up to 400MHz. It's only really IRIX and Cray stuff that goes over 64 processors, although you could argue that Beowulf can go that high, but that's clustering, not SMP.
Journaling would be good, as it would dramatically improve the reliability of linux under system crashes (rare under linux) or other hardware failures (isn't a lot you can do if the power supply dies). Bear in mind, however, that Solaris has only just put this in by default; before you had to pay for things like Solstice disk suite to get a journaling filesystem, so linux doesn't lag that far behind.
--
Anything written ambiguously and/or poorly is FUD wether it was meant to be or not. FUD ; Fear,Uncertainty and Doubt. Ambiguous writing seems to cause Fear, Uncertainty or Doubt in the minds of those who read it seeking objective information.
In which case, the complaint would be justified.
Solaris has auditing and ACL's for 'Trusted Solaris', though I have not (yet) used it. This is how one obtains B1 security for Solaris.
VMS has done auditing correctly for a long time. It has made my job much, much easier.
Click here for Sec urity Event Classes that can be audited/alarmed. (It wouldn't hurt to read the whole Security auditing section. The ANALYZE/AUDIT tool is very nice.)
It is also good to have hardware-level events logged. A couple of years ago I had a VAX crash. I simply did ANALYZE/ERROR/SINCE=TODAY and found out I had a SIMM that was having ECC/parity errors. Since it logged the bank of memory with the failure I knew exactly which SIMM. I called the DEC service guy, he came out, and we switched it out during lunch. No one noticed.
Yes, Linux could use these things.
The Research has couple of flaws - some the fault of DH Brown others just due to bad timing.
The Features are poorly tested and weighted badly - Its all very well saying something has a feature but if it is Shite then it is of little use.
NT has many applications and a variety are included (when you pay big bucks) - unfortunately they are usually unstable due to the nature of win32 and the instability of the NT kernal.
The research weighed in favour of what NT had and badly in what it hadn't and made it worse by not testing or benchmarking them.
Linux should of came in at least equal or even ahead of NT - adding SP4 and the extras is no different to adding the 2.2 kernal (my experience leads me to believe the latter is easier and quicker).
the other problem is bad timing - the research pre-empted many important events such as SAP and other major Enterprise Application vendors not only supporting Linux but supporting it with SMP and other Enterprise features.
The research also pre-empted the 2.2 kernal release which is a shame.
Intranet/Internet Developer & Linux Advocate
Not *exactly* journaling, but one project is here.
The consensus was that the article was talking about "many" processors meaning more than two or four... does Linux run on 128 processors?
In terms of "logging" people felt that this was referring to a journaling filesystem. Granted, I don't know what this means. :-)
I'll admit, though, that the way it's written, it certainly does look a bit like FUD.
Linus also didn't plan on supporting non PC hardware or non x86 chip sets.
There's nothing stopping something like this being a compile option or a patch set, etc. Though I don't know anyone who's run into this 'limitation' yet either.
--- http://foo.ca
Journaled Filesystem. Linux needs it to be competitive with the big dogs. If the feature were available it would remove a huge barrier to Linux's success. The other OSs mentioned all have this feature (HP-UX?).
I guess Beowulf and syslogd are just random collections of letters which don't mean anything.
:)
... and missing.
Well, why cant they be? After all, thats how MS puts out its code. An infinate amount of monkies pounding out source
The rest of the article isnt even worth flaming its so wrong.
---------------------------------------
The art of flying is throwing yourself at the ground...
Since we all know what classic FUD is, this report shows what diet FUD would look like. Does this mean that one of the two celeron processors on my box (which I painfully assembled after 3 days of soldiering) is not working SMP? Exectly what kernel did these people evaluate? Linux 1.0x?
I'm really suprised a report like this could be posted on a non-biased (ahem) media like MSNBC. Or has the MS part of MSNBC finally eaten up all good reporting? This is seriously sick. Along the line of linux not supporting SCSI, Ethernet, and kitchen sink networking.
--
If I'm not mistaken, evertime an article is reposted on a popular network, it is editor read and corrected before being posted. (If this is the case, why did MSNBC let this article into their servers?).
--
Spends most of the article reluctantly saying good things about Linux and then the one piece of "evidence" used to support the thesis is totaly, utterly false. Someone was dredging the bottom of the FUD barrel for that one.
Wow. We're being evaluated according to high-end
criteria! This is *good* news. So we're not
doing so well against supercomputers! Hooray!
It sounds like we beat windows a long time ago.
-fb Everything not expressly forbidden is now mandatory.
He's been saying this for a while.
Why doesn't he just put the working code out
there in a cvs repository. What's all this
"waiting for a preliminary release?"
I'm scratching my head as I walk through this
bazaar...
-fb Everything not expressly forbidden is now mandatory.
Lea
my work might not be so good, but at least I take credit for it!
and it says this "even though I know very little about Linux, I /do/ know that there IS SMP and logging. a lot of it. and this article says there isn't."
/critisize/ something, but denying that it exists is another thing entirely. I don't like the whole system of grades here (Berkeley), let us say for the sake of argument. does that mean I can deny their existance? NO! (but sometimes I certainly wish I could...)
just plain wrong. I'm not even a Linux advocate, but I do like it when people could MANAGE to get their facts slightly straight. For example: one may
the issue for some is the issue of quality of these things (or so it seems. I know squat about it). the issue for just about everyone should be basic correctness before discussion. basing your arguments on articles like these may seem to give you ammunition, but all it does is spread yet another lie, and annoy the people who DO know what's going on (which, as I'll say again, is not really me too much)
Lea
I sent a comment and the letter bounced.
Appararently the address for EMAIL is
a bit bucket.
You can tell that they didn't do any research whatsoever. I bet they just typed it up on MS Word and submitted the story. These guys are on crack.
I guess Beowulf and syslogd are just random collections of letters which don't mean anything.
Right, Linus said he dosen't plan on it... but he also mentioned that scaling above 16 processors would require mods that I'm sure could be made into a patch of some sort. MPLinux? A small group of SMP people that follow the latest kernel and apply minor patches to bring it up to par? Couldn't be that difficult =)
Maybe it's possible to send off an anti-FUD piece to another publication about the study and its flaws; then inviting them to report the study,m and then say that it is factually innaccurate ... (-; The journalists are hungry for stories on linux, and hungry to have a poke at a competitor ...
But while the report at least basess its attacks on the truth, the article resorts to outright falsehoods. One wonders how many falsehoods of this nature that we see in the press would constitute actionable material if they were made about a commercial OS ...
- Reliability The report more or less dismisses this, and lauds NT's reliability features ( such as journalling ) without verifying that the features actually lead to a system that is more reliable in day to day use.
- De emphasising or ignoring NT's shortcomings Severe shortcomings of NT's out of the box setup , such as nonexistent security, absence of disk quotas, etc seem more or less ignored. By contrast, the linux they assess seems to be relatively "out of the box" ( for example, SMP support. They don't really consider 2.2 here. ) And while they are quick to dismiss linux as a "serious" OS , they seem to acknowledge NT as enterprise ready despite its lack of essential features ( such as remote administration, disk quotas, etc)
cheers,--
=================================================
Was that Linux was being compared to Big Iron and found lacking. Linux logging is no worse (and often better) than most commercial Unices, but the only place I've seen absurd levels of multi-processor and system logging are in the "Real Computer" world.
===============================================
With the exception of Compaq Tru64 UNIX (aka Digital UNIX). Syslog is syslog, on just about any UNIX, but Tru64 has something call the Binary Event Log (binlogd). This logs a whole host of hardware events and some software, with event types, severity level, and device register information for analysis. i.e.
Hardware-Related Events
~~~~~~~~~~~~~~~~~~~~~~~
100 CPU machine checks and exceptions
101 Memory
102 Disks
103 Tapes
104 Device controllers
105 Adapters
106 Buses
107 Stray interrupts
108 Console events
109 Stack dumps
199 SCSI CAM events
Software-Detected Events
~~~~~~~~~~~~~~~~~~~~~~~~
201 CI port-to-port driver events
202 System communications services events
Informational ASCII Messages
~~~~~~~~~~~~~~~~~~~~~~~~~~~~
250 Generic ASCII informational messages
Operational Events
~~~~~~~~~~~~~~~~~~
300 ASCII startup messages
301 ASCII shutdown messages
302 Panic messages
310 Timestamp
350 Diagnostic status messages
351 Repair and maintenance messages
You can specify the following severity levels:
severe - Specifies events that cannot be recovered and that are usually
fatal to system operation.
high - Specifies events that either can be recovered or cannot be recovered
but are not fatal to system operation.
low - Specifies informational messages.
This information is like liquid gold when it comes to pinning down hardware problems. I sorely wish Linux had something similar.
Macka
I have heard that support for a journaled filesystem for linux has been started. Does anyone have any more information on this?
Use the above URL to comment to wsj---please make sure you reference the URL of the article on msnbc which has their name on it, and the specific factual errors.
Actually, mainframe CPUs themselves are not that much more overwhelmingly powerful then desktop CPUs. The mainframe benefits mainly from I/O channels, block mode devices (including terminals), and better scheduling algorithms.
... as opposed to receiving an interrupt when each block transfer completes.
I/O channels are CPUs dedicated to supervising I/O operations, so that if you need to read in 100 blocks from disk, you build a list of the blocks along with their addresses, and start the channel. The transfer is done, and you receive an interrupt when it is complete
This really pays off when you include block mode terminals, like 3270s. A 3270 contains a screen buffer. When you are working in a text editor like VM XEDIT, everything you type is stored in the terminal until you press return, or a function key. At that point, the terminal transmits a list of all the screen fields that have been changed. If you were running VI on a unix system, you would be peppering the computer with console interrupts with each keystroke. More if you are running X. This is how mainframes can efficiently support 1000+ online terminal users.
Mainframe scheduling algorithms are specifically designed to separate the workload into interactive and non-interactive users. If the scheduler decids that you are an interactive user, you get small timeslices and more of them. When you start your big program, and it goes CPU bound, the scheduler notices that you are using your full timeslice, and quickly moves you into a different queue, so that, for instance, after a while, your program will receive a timeslice that is 16 times as long, but only receive the timeslice 1/16 as often as an interactive user. So your background process lurches along, but you don't notice because the instant it starts doing I/O to the terminal, it becomes an interactive process again.
These sorts of tricks are what keep mainframes from appearing to be "bogged down" even when their resources are massively overcommitted.
These algorithms have been fine tuned for about 30 years, and are specifically designed to best utilize block mode I/O devices, and large numbers of interactive users attached to boring 3270 terminals. It's a VERY different workload then you'd find on a Unix system, and the two workloads don't compare well.
In fact, one of the biggest problems with mainframes is running TCPIP efficiently, because TCPIP *does* pepper the system with interrupts.
- jms
NT can act as a router, but is not so by default. Setup is simlar to unix, with a ROUTE command. (Or MS has a no cost routing add-on with a GUI.)
NT does broadcast all the time, which seems to be a characteristic of the NetBIOS-over-TCP/IP protocol. Is OS/2 or Samba any better in this regard? Need to find someone with a network monitor to check
--
Business. Numbers. Money. People. Computer World.
> I guess that 8 processor system I got running RH 5.2 on last week didn't count in the study.
Nope, it doesn't. Get it running on 64 or 256 processors with true shared memory and then it might. Beowulf doesn't count either, that's not SMP it's clustering of independant machines.
Linux is cool. It rocks the medium-low end of the computer world and nukes NT. But, it can't touch the high end stuff - yet.
"Among them are the ability to run simultaneously on many processors in a single computer and to keep a log of what the computer has done"
Fact:
The largest (talking processors) NT server you can get is a 8 processor system from NEC/Zenith/Bull, HP had a 8 processor NT server to, but gave it up. So there are linux boxes with two times as many processors as the largest NT box.
If "keep a log of what the computer has done" is a referring to a journal file system (I have been working with AIX and jfs for years) and it was not clear to me. Then it is true, linux don't have this now (I have heard it is worked on?). But NT don't have this either!
The article may not be intentional FUD but it certainly comes close. Don't expect to see a lot more of articles that bash linux all the way, instead we are in for a lot of "Linux is quite good, but note quite there" FUD. (the same they did to OS/2 I have heard)
Linux, coming to a desktop near you!
cjs
The world's most portable OS: http://www.netbsd.org.
Ok, so, this article possibly referenced the 'study' done a while ago that placed AIX first place. So, with my trusty rs/6000 to my left and my linux box to the right, let's compare logging:
:)
AIX: Logging Sucks.
Linux: Logging Rules.
NT: NT does logging?
well, there we have it, a fairly comprehensive examination of the logging capabilites of each.
Of course, there's no arguing about SMP on rs/6k's.. it's awesome. But, with the way things are going around here, we can expect to see IBM SMP techonology go to linux loooong before it goes to NT.
Go big Blue!
i browse at -1 because they're funnier than you are.
hi all.. for clarity, i'd like to reiterate what others have said early on in these discussions... the most probable interpretation of the wsj article author's statment is that linux (2.x) currently lacks a log-based (i.e. journaling) filesystem. (such as the original veritas (vxfs) filesystem, etc.) just figured i've seen so many people who didn't seem to catch that (very salient) point, that it needed reiterated... Peter
It sounds like the reporter misunderstood what was being said (no great surprise from a rag like the WSJ.)
/could/ mean that NT's application support outweighs it's huge technical shortcomings, and that until Linux is as much better than NT as are the commercial Unices, it isn't really an option.
By multiprocessing, I'm positive that they mean SMP with more than four processors. I don't have any experience with Real Computers (my largest was a 16 processor Power Challenge) but I imagine that the big iron goes way up past 4 Xeons. This seems more like a hardware deficiency than a Linux-specific one.
And by logging, what was pretty obviously meant was the kinds of system accounting that mainframes provide -- a tremendously neurotic cascade of information that only a tremblingly uptight, card-punching MIS monkey could love. In this respect, Linux is no better nor worse than any commercial Unix, and, due to the open nature of the code base, better than most.
Or perhaps the article meant "logging" as in a log-structured filesystem (aka a journaling fs.) This is a real minus for Linux, and could concievably keep it out of the real enterprise, where it just can't compete with the Big Boys.
I even think I understand the inclusion of NT as a superior alternative (not, mind, that I agree.) The author
Don't forget that what the datacenter wants out of it's application platform is very different than what makes a good web|SMB|login server. Which doesn't excuse nor explain why people are buying NT, but that's corporate evolution in action, as far as I'm concerned.
Regards,
JFB
To spur "enterprise Linux," Big Bang, the distributed two-phase commit.
It's true that Linux doesn't scale like Solaris on the Big Iron to 128+ processors. On the other hand, neither does NT, and NT was ranked in front of Linux. I know that VA research demoed an 8 CPU Xeon system at Linux Expo, and I've heard about someone running linux on a 12 CPU Sun system. Can NT even do that? Why on earth would NT be grouped with the Big Iron OSes like Solaris, Irix, Aix, etc?
As for the journaling file system, I think that that's on its way, though I don't know for sure.
They laughed at Einstein. They laughed at the Wright Brothers. But they also laughed at Bozo the Clown. -- C. Sagan
As far as journaling file systems, it's my understnading that that's on its way, but it definitely isn't here yet.
:-) I know that I don't have the year 2038 problem, nor do I have the 2(4?)Gig RAM limit on my Alpha.
As far as SMP systems, ask VA research how their 8-CPU Xeon system runs. Care to comment, Chris?
As for terabyte files, try an Alpha, or any other 64-bit platform. I'm fairly sure that my Alpha could do terrabyte files, if I only had the hard drive for it...
Now, can NT do 6-8 CPUs worth a damn? I'm fairly certain that NT can't do 64+ CPUs worth anything. And does it have a journaling file system worth mentioning? I've never really dealt with journaling file systems. Does anyone know (btw, worth a damn/worth mentioning means on the same sort of caliber as Solaris/Irix/Aix/etc. can do it)?
They laughed at Einstein. They laughed at the Wright Brothers. But they also laughed at Bozo the Clown. -- C. Sagan
Someone please tell me then what my SMP linux box has been doing for the past year or so??? and why does it tell me I have 2 processors, and why does it tell me which processor is running what process? (with a kernel patch)
is this guy a total M$ idiot or is he just pretending?
next he'll be telling us that UNIX is a new operating system, and that M$ has ben around for 100 years...
this guy is a total dips---
Only 'flamers' flame!
and guess what... you cant do most of that on NT either
Only 'flamers' flame!
"...and it's inability to keep a log of what the computer has done."
If this paraphrase of the direct quote given in the article is indeed correct, in all it's false glory, then I think those people you know should have worked a little harder.
If you honestly think that Linux equals the best commercial software, you really aren't looking very hard. It's a shame that I've seen so many similar (and equally uninformed) comments on this thread.
Cheers,
ZicoKnows@hotmail.com
"Linux is only free if your time has no value" -- JWZ, mozilla.org
OK, first of all, when the report was done, both distributions used 2.0.x kernels. So there's one reason for low SMP scores. I won't delve into all the people around here who used to tell us how wonderful SMP was back in the 2.0.x days, and are now telling us how wonderful SMP is in 2.2.x.
Fact is, while the SMP is improved, it's still not there yet, and many people (people actually using the stuff for business, not all the uninformed zealots that you see ranting about MSNBC in this thread) have doubts that it's ready for the enterprise.
As for the SMP questions, they're legit. There are major functions within the Linux kernel which are still, even with 2.2.x, not reentrant. True asynchronous I/O is missing here, too, another strike against it. These things aren't really going to effect most people here in a big way, but they are areas that need to be improved (along with others) for more people to start seeing Linux as enterprise-ready -- and no, I don't think this report was aimed at the usual Slashdot audience: I imagine that the intended audience could give a rat's ass about KDE vs. GNOME, xamp, or themes.
Cheers,
ZicoKnows@hotmail.com
Well at least the summary, the real thing costs $995. There are some interesting items:
"-based on their functional capabilities as of April 9, 1999."
?? It is from the future?
"Kernel 2.2.5 used by OpenLinux 2.2 should improve it's SMP scalability somewhat."
So they did use Kernel 2.2, but what do they mean by should? We read on and see:
"Little industry-standard or even propretary benchmark evidance has emerged that demonstrates the performance improvements of database or Web server application running on SMP systems under any Linux distribution."
So they didn't actually test this themselves? They go on to show some charts without Linux results, showing instead "No Published Results".
The report does make several very valid points discussed already which will hopefully be addressed in the future.
The best thing said about Linux is System Management. They really like Linuxconf shipped with Redhat, especially the many different interfaces available, "command-line, text menu, X/Motif, and Web-browser." Although, there isn't a Motif interface on Linuxconf is there? I never paid attention. They pan NT here.
They make several valid criticisms, everyone should read the summary.
*Note: any spelling errors in the quotes are my own, I just typed them out, I didn't cut and paste.
Q.
Pardon my ignorance, but could someone please explain what a "journaled filesystem" is? I have not heard this term before.- ----
-----------------------------------------------
Jamin Philip Gray
jgray@writeme.com
http://students.cec.wustl.edu/~jpg2/
Celebrate the finer things in life
-----Original Message-----
From: dhbrown.com
Sent: Tuesday, April 06, 1999 1:02 PM
To: tcooper
Subject: Re: Feedback
Tom,
The report carefully defines its terms and conclusions,
summarized briefly in a free summary on our website at
www.dhbrown.com.
The news report you cited is indeed somewhat inaccurate
(it is a paraphrase rather than a quote). Linux
does "run simultaneously on many processors" if
many equals, say 4 to 14. Our report however does
also look at systems that have proven performance in
both production and industry-standard benchmarks up
to 64-way SMP. Linux has not yet reached that level.
In fact, we were mildly shocked that, despite reading
various kernel lists and talking to various Linux
vendors, there is currently no really good publicly
verifyable benchmark evidence of Linux's scalability
even on 2-way or 4-way workloads (although I would be
surprised if we didn't see some in 6 months). It boots
on a 14-way as David Miller demonstrates, but the
scalability claims are still a little in the air.
I wouldn't make a claim that Linux doesn't scale.
I would make the claim that Linux advocates have yet
to reasonably demonstrate that it does.
"Keeping a log" is a similar over-simplification. I believe
the reference was to "event management" facilities just
now becoming available in UNIX where all the system
logs can be accessed from a single console in a consistent
manner. Managing various logs in UNIX has long been
more painful than it needs to be.
Thanks for your respectful feedback.
Regards,
Greg Weiss
Research Analyst, Systems Software
D.H. Brown Associates
tcooper on 04/06/99 11:41:00 AM
To: DHBA Systemsw/DHBA
cc: tom_cooper@bigfoot.com
Subject: Feedback
According to http://www.msnbc.com/news/256197.asp Your study claims
that
"Linux currently lacks some of the features demanded by corporations
that intend to run their entire business on computers. Among them are
the ability to run simultaneously on many processors in a single
computer and to keep a log of what the computer has done."
What sources can you cite for this assertion? Linux is multi-processor
scalable, and does provide logs that are at least as detailed as
anything that you can retrieve from an NT box.
Respectfully,
Tom Cooper
But Herr Heisenberg, how does the electron know when I'm looking?
Linux fares better in more specialized tests, though. In one recent test by Smart Reseller, a trade publication, Linux proved to be more than 2 1/2 times faster than Windows NT in certain common computer-networking tasks.
That is utter bull! The Smart Reseller test was the STANDARD ZD SMB/WINDOWS SERVER BENCHMARK!!! It is how ALL WINDOWS SERVERS ARE BENCHMARKED! It IS THE STANDARD OF WINDOWS NETWORKING BENCHMARKS!
Just more FUD.
-- Bryan "TheBS" Smith
Independent Author, Consultant and Trainer
I read the WSJ article, and it was clear to me that the study seemed to have suffered a little bit in the translation (having been dumbed down).
So, I decided to go directly to D H Brown's website and hear it from the horse's mouth. However, they want you to pay the meager sum of $995 (that's nine hundred) for the privledge of reading it.
Sorry, but any 'study' that is done clearly for profit is by definition unscientific and almost certainly biased. And it is no surprise that the 'study' seems to come out wholly in favor of companies who practice the same insane pricing schemes.
Praise the Force Field! Praise the Laser Project! Slackware Loon #19830573
I guess researching a story with more than one source is just something you have to remember for a test.
Geez!!!!
I was noticing that Lee Gomes, the writer of this artical, is also the Moderator for the afore mentioned AIIM '99 panel, http://slashdot.org/article.pl?sid=99/04/06/122324 5 think it might be baised a bit?
End Transmission....
Well, Linus has said that he doesn't plan on making Linux scale above 16 processors, because after that you're hurting people who only have one.
This is posted at msnbc.com. Any questions?
feathres demanded by corporations
The first problem: "Run simultanously on many processors" has been answered by SMP.
The second problem: "Keep a log of what the computer has done." Needs some clarifications. If they want process accounting, here's a bunch of Accounting packages for Debian (Yes, you can find them in RPM format too). If you want to know _exactly_ what the computer is doing... Look at the source. Try that with IBM's Unix :).
Go look at page 40 of the 3/29/99 issue of PC Week. It says the exact same thing. Kinda strange how MSNBC, WSJ, et al. all say the same thing. It's amazing how easy it is for a falsehood like this to get blindly written up as fact.
Looks like someone at the AP had better start doing some research lest they get accused of being FUD mongers...
-B
Ash and Hickory, straight-grained and true, make excellent bludgeons, dandy for the cudgeling of vegetarians.
1. Maybe they are talking about file system journaling. Maybe they are talking about extensive file (and object) access logging for auditing purposes. Either way, Linux doesn't have it.
2. The latest WinNT magazine has a good article compairing the NT and the Linux Kernel. It may be biased, and it may be making mountains out of molehills, but it does contain a reasoned and supported compairison of the two OSs and it would be good to have a linux person post an equally well supported rebuttal.
Well, I went to look at the report, but it is
$1000. But even without reading it, how can the
statements about multiple processors and logging
be right?
where did Stu mention NT?
The point was that the article was pointing out Linux's inferior scaling and logging capabilities when compared to commercial Unix brands like Solaris and AIX. However, it didn't bother to mention that NT has these same falacies (and more), thereby implying that NT was better than Linux.
Mr. S. Tweedie is working on such a project. He has indicated in the kernel mailing list that a preliminary version of fjs would be available in about 4 weeks or so. If you have time and resources, help him to iron things out.
You forget to mention that such a database can be constructed on Linux/Oracle. If you want a single table to be one TB (which is unlikely as a dataware house consists mostly of MANY tables), just create a tablespace of >500 datafiles, each 2G. Being a data analyst at a large telecoms company, the largest,single daily file that I run queries against is about one Gig (a few million calls) or so. Performance reasons usually keep datafiles small and spread across multiple disks.
If I remember correctly, a report appeared on the kernel mailing list stating the successful installation of linux 2.2.X on a 16 cpu Sun box.
And about the 2G file size limit, I don't think this is the limit for Alpha and other true 64 bit boxes.
I would be reasonably comfortable in saying that slashdot is a biased source of information too
Slashdot does not pretend to be a journalistic site that provides impartial reporting. The opinions you see on slashdot are very obviously those of individuals. MSNBC purports to provide unbiased information. That is the distinction. If a Time/Warner media outlet is biased towards Windows NT, we might be able to effect a change in that, but if an MS-owned media outlet is biased, there is little likelihood of change.
Mike
--
Mike
--
"Wi nøt trei a høliday in Sweden this yër?"
OK,
maybe they worked ward on the report. But, the report is still wrong (some people are incompetent at their job, you know...).
First of all, accton is a very complete logging utlitie, but it lacks a GUI. Most commercial UNICES have a more user-friendly interface.
Also, these guys probably havenM't heard of smp support for Linux. This is pure FUD.
Papi
- Chernobyl used windows
OK,
first of all, I've worked with solaris on sparc stations, and I'm not very impressed. A couple of my friends that use Linux at home call the OS slowlaris. For some reason, it has a very slow X (but, it never crashes).
Solaris of course offers better support and maybe the best kind of on-line documentation, but the interface is ugly, and the OS is slow.
Also, Linux beeing behind windows-nt is ridiculous. First of all, win-nt doesn't offer multiuser facilities. It broadcasts too often and they're is no way to know why it does so, and they're is no way to change it's behavior. It just keeps on eating the bandwidth for no reason and denies real packets the chance to go where they should.
It is also a router by default. OK, it is easier to configure as a router, but just imagine a sales person comming in, plugging it's computer to demonstrate a piece of software (using an available or even a used IP). Then, the win-nt broadcasts thet it is now routing all packets, and, your whole network goes to the bit-bucket.
Papi
- Chernobyl used windows
Hmm,
maybe one of them is a little clueless. But, ones worked as a sys-admin for 6 years beofre he decided he was more into programming, and the other was a sys-admin on an sgi router for a while (part time).
As for me, I do not consider my self clueless in UNIX. It is just that I don't feel like staring at a "welcome to solaris" (written in a wide array of languages) screen for 3 minutes.
Also, I'd use themes with olwm, it's just that sun using risc processor, and all the executables in my account are huge (bash takes a little over 3M stripped). Also, the only uglier window managers I am aware of (and I used quite a few) are mwm antd twm (maybe fvmw, but, it's a close one).
You must also consider that it costs over 12 000$ a year to maintain a network of sparc stations. This money could easilly be used to upgrade a fwe PCs that could run under Linux.
Of course, sun's computers are more powerfull that PCs, but, they cost so much and new PCs are becoming more and more powerfull, that after a year or two, you just realize that you would have been better off buying a PC in the first place.
Papi
- Chernobyl used windows
Maybe he has no e-mail address, or doesn't know how to use it...
:-)
ms
i tried to email wsj with some facts. i used the email addy linked at the top of the article. it bounced back to me faster than a superball.
joe
hee hee hee monkeys hee hee hee
that image makes me smile.
something that microsoft's software doesn't do.
hey did anyone ever think of this? n't is how you usually shorten not in a contraction. leave out the apostrophe and you have nt! windows not. or better yet windows not server and windows not workstation. wow, what a descriptive name!
joe
I've been following the MSJ and MSNBC reports on Linux for some time out of shear(sp?) perversity and simple amusement.
Take it from whence it comes... even the good stuff and there has been quite a bit of good stuff on Linux from these sources.
Meaning? Not all FUD! But all done for MS's benefit. Make Linux look good so it can be competition to keep the DoJ at bay as well as spur some of the Redmondites to better/faster coding. Then make it look not as good and MS look better. It's a dangerous gamble but I think it's the only one MS can make given the supposed fear or the 'garage factor' Bill might really have.
If what I said is nonsense,
I'm making a point with it.
If what I said makes perfect sense,
you obviously missed the point.
I guess that 8 processor system I got running RH 5.2 on last week didn't count in the study. Neither did any of the quad-xeon systems I've installed it on (or VA, for that matter), or the dual PPro I have sitting next to me.
Who are these people and how far up Gate's butt are they?
der dee der.
The reason why mainframes need massive cpu power is not because of the logging, its because they have a TON a power all packed together that is needed in certain applications. One person at my work commented that pcs are great for word processessing but if you have to do any number cruching they just get bogged down (and in the case of a windows pc, crash.. I mean it). Meanwhile, the same number crunching can just be thrown onto a mainframe and it'll get done. You don't have to worry wether or not it will be done at all.
-?-
/*AIX: Logging Sucks. :)
Linux: Logging Rules.
NT: NT does logging?
*/
I agree with the AIX part
Nothing like using the "errpt" tool to ruin your day. or SMIT, ick.
Lowmag.net
Since most people who administer commercial Unix boxen don't enable them, many people don't even realize that some systems have rather verbose extensive logging mechanisms. The best that is out there is the Sun Basic Security Module (BSM) audit facility. It'll generate lots of logs and sure it takes a fair bit of resources not to mention disk space, but it allows you to run fairly sophisticated host-based intrusion detection systems and very good post-mortems!
Because of Linux's open nature, it would be very useful to have a verbose audit trail mechanism. This would allow security researchers like myself to base new systems on Linux more easily. (and yes, it would most likely end up being GPL'd!)
A journaling file system would be super neato as well. From a security standpoint, one could get a much better idea of what an attacker did to the system files even without running tripwire.
In closing, these are somewhat advanced features and there is no reason why they can not be added to Linux. I believe most of the commercial Unices had them added to an existing system as well. Well, Nuff said.
I especially liked the "sponsored by microsoft" tag that appeared on the byline.
Gray lady? Isn't that the NY Times?
.com that's adopting it for mission-critical service (Salon comes to mind). Good news offsets the bad, or at least muddies the water enough that the decicion makers will have to consider all the evidence.
I don't think decisions to adopt or avoid Linux will be made solely on the basis of this MSNBC report, the original WSJ story, the free executive summary, or on the $995 report. My impression is that word-of-mouth experience drives Linux adoption as much as any media reports. Besides, for every report that says you can't use a Linux box to replace Big Iron, there's another one reporting some well-known
The man who does not read good books has no advantage over the man who cannot read them. - Mark Twain
Go to this Infoworld article for a clearer version of the story. (Yes, they are talking about journaled filesystems.)
they say so in the survey review pdf that u can download.. oh and if you wanna read the whole thing you have to give the guy $995
unless you have $995 to spend. they dont give the survey, they sell it
I don't understand the number of criticisms I've seen over the past few months saying that Linux doesn't support multiple processors.
The only thing I can think of is that distributions generally do not install an SMP kernel by default, but make it an additional option. Is the person who researched these reports capable of changing a base installed config.
This means that the core image of a process is kept on disk, so if the processor fails, then it migrates to another one and carries on. This is cool stuff. We have had processors melt down on Tandem's and they *still* keep going. You shut it down sometime later, and replace the board. Meantime, your mission critical system just takes a (np-1)/np performance hit.
The top end Sun machines have this sort of capability now ( I think ), but I'm not sure the Linux SMP can do these kind of wonders yet.
That's not to say that it can't, just that it doesn't. And until it does, its unlikely that people will be betting their ATM networks on Linux. They'll be sticking with their Sun's, Tru64s and Tandems.
Not that WindowsNT is much of a competitor in this market anyway.
I would, of course, be happy to work on this if someone is willing to donate a 6+ processor E4500 to play with
Was that Linux was being compared to Big Iron and found lacking. Linux logging is no worse (and often better) than most commercial Unices, but the only place I've seen absurd levels of multi-processor and system logging are in the "Real Computer" world.
I think the writer's opinion seems somewhat biased (surprise, surprise) but he brings out some reasonable questions: just how far can a single Linux box scale, and for what tasks?
Now we should get out there and improve all the little things that need improvement to help Linux, and *nixes in general, reach these entrenched Heavy Hardware markets.
Complete aside: I believe the reason mainframes need massive CPU power has nothing to do with capacity and everything to do with the tremendous overhead of the monitoring, accounting, tracking, logging, and process/resource management features of most Real OSes (e.g. OS/360).
I skimmed this article at a local news stand. It claimed [very much paraphrased, consume with a Tbsp of salt] that Linux couldn't be considered an enterprise quality OS because:
IMHO, this article was lame because its conclusions were drawn without any benchmark evidence. Just because certain aspects of the Linux kernel aren't as elegant and buzzword compliant as they could be doesn't mean it does kick ass in real production use. here's my Linux mailserver's uptime: 12:08am up 193 days, 16:28, 1 user, load average: 0.05, 0.03, 0.00
the only reason its not longer is it wasn't connected to a UPS. The only time I had to reboot before that was when my company changed ISPs.
"Yeah well
I skimmed this article at a local news stand. It claimed [very much paraphrased, consume with a Tbsp of salt] that Linux couldn't be considered an enterprise quality OS because:
IMHO, this article was lame because its conclusions were drawn without any benchmark evidence. Just because certain aspects of the Linux kernel aren't as elegant and buzzword compliant as they could be doesn't mean it doesn't kick ass in real production use.
here's the uptime for my Linux mailserver which supports six offices:
12:08am up 193 days, 16:28, 1 user, load average: 0.05, 0.03, 0.00
the only reason its not longer is it wasn't connected to a UPS. The only time I had to reboot before that was when my company changed ISPs.
"Yeah well
he seemed like a pretty nice guy over the phone; a little confused, but he joked around and stuff with me. he wasn't FUD-controlled or anything -- just uneducated. (i mean, he called ME -- that's as good a sign as any ...)
Hey, isn't a guy from DH Brown going to be in that panel with the MS Drone, Linus, and Maddog?
It could be interesting, but I don't know. Linus and Maddog are engineers (I think Maddog is, anyways), while the other two are not. Somehow, I doubt they'll be speaking the same language. (This happens a lot when people whose opinions on some issues are so different.)
You're a suburbanite.
This article is almost identical to to the articles I have read in both NT Magazine and on the MSDN site.
The study said that stability evidence was almost entirely anecdotal... Anytime I have had a serious problem with NT, Microsoft says that it is anecdotal and I must be doing something wrong.
This article may be correct, but it is FUD.
I don't understand the number of criticisms I've seen over the past few months saying that Linux doesn't support multiple processors. The frequency of such comments suggests to me that either there is such an issue with linux (which I'm pretty sure isn't true) or someone is "spreading rumors" to media sources about linux and it's capabilities. (an alternative I find just about as unlikely). Can anybody suggest what might be causing all of these complaints about SMP on Linux?
I know that this ended up on the MSNBC website, perhaps for obvious reasons, but it says that the article came from the Wall Street Journal. Does anybody know if there's an MS connection to the WSJ?
I don't mean to criticize, but perhaps this is a case of the pot calling the kettle black. I would be reasonably comfortable in saying that slashdot is a biased source of information too. (albeit a much better one because of the possibility of many many opinions being posted) But seriously, when was the last comment you've seen talking about the (admittedly few) good things about windows?
"I don't mean to criticize, but perhaps this is a case of the pot calling the kettle black. I would be reasonably comfortable in saying that slashdot is a biased source"
How do you define a biased source? Is a humor site biased because it provides humourous articles? Slashdot has NEVER been a source of FUD against Microsoft (I stand corrected!). Unfortunately Microsoft seems to use their own controlled press for spreading FUD against linux. So that's free speech I guess.....
"This article is automatically bad just because it's posted on MSNBC."
C'mon folks. We have to do better than that.
The article is subtly written. Such phrases as
"...inferior to commercial operating systems, such as Microsoft Corp.'s Windows NT" on the large slug at the top, and "Linux will begin moving to the desktop world dominated by Microsoft" give a hint as to which side the article's already on. Try to ponder if the writing would have stayed the same if there was no "MS" in "MSNBC".
The other operating systems which Linux is apparently inferior to aren't even mentioned until the last two paragraphs. There's a definite Microsoft slant in this article, and it's sorely unfounded.
The following are from an email from Greg Weiss (grweiss@dhbrown.com) :
The news report you cited is indeed somewhat inaccurate (it is a paraphrase rather than a quote). Linux does "run simultaneously on many processors" if many equals, say 4 to 14. (me: but they are also looking at up to 64 SMP, and linux hasn't reached that level)
In fact, we were mildly shocked that, despite reading various kernel lists and talking to various Linux vendors, there is currently no really good publicly verifyable benchmark evidence of Linux's scalability even on 2-way or 4-way workloads (although I would be surprised if we didn't see some in 6 months)
I wouldn't make a claim that Linux doesn't scale. I would make the claim that Linux advocates have yet to reasonably demonstrate that it does.
"Keeping a log" is a similar over-simplification. I believe the reference was to "event management" facilities
So, If you have Linux running on a SMP system, PLEASE mail Greg Weiss (grweiss@dhbrown.com) and tell him!
NT's "NTFS" and IBM's AIX "JFS" are both journaled. I don't think the HP-UX filesystem is, but I am uninformed.
I do know that both AIX and HP-UX have nifty logical volume support, so you can add space to any logical partition from any *physical* partition at any time. I think that's probably an 'enterprise' requirement much more than a sort of worstation thing.
What other filesystems are journaled?
--
: tedd
So write it!
The report read that Linux has bad or worse SMP than the rest. But we knew this! The 2.0.36 kernel that comes with Red Hat 5.2 doesn't do SMP very well. No surprises here, no arguments. But I'd like to see next year's report, when they analyze Linux with a distro using the 2.2.x kernel. Then we'll see.
But you can still get the 16 page summary in PDF just for registering. Look on the site harder.
#3 - Stephen Tweedie posted to linux-kernel recently that he'll have testable journalling code for ext2 in a few weeks.
#4 - check out the "fake" utility
#6 - get a 64-bit platform
#7 - get glibc 2.1 (32-bit UID's)
Another article pointing out Linux shortcomings in contrast to NT appears in April issue of NT Magazine (print version). The magazine will not post the article to its web site until July. It is authored by:
Mark Russinovich, PhD in computer engineering from Carnegie Mellon University. Russinovich has authored a variety of NT focused articles.
mark@systinternals.com
http://www.sysinternals.com
There are some similarities in the articles. Russinovich indicates Linux will continue to evolve and these shortcomings will disappear.
[I've based this post on the PDF document someone, I've forgotten who, kindly posted on this board - not on the whole report. I don't have $995 to spare, funnily enough.]
;)
Considering the amount of hype Linux has received in the past few months, it is only to be expected that a report such as this would be forthcoming. I think it is important to bear in mind that while the report does not truly compare like for like, it is probably necessary that a report such as this _is_ produced before the general populace start expecting too much of the Linux community, and when they are disappointed, turn away from you never to return.
The report does bring to light a number of reasons why I and other other sysadmins I know have generally steered clear of Linux in favour of *BSD and commercial UNIX systems - and, when the occasion has demanded it, Windows NT. I know this may seem like blasphemy to many readers, but corporate necessity wins over any prejudices or principles.
Pricing: I don't think anyone will argue about this. Linux is, in fact, cheaper than all the others.
Scalability: Linux is not as scalable as operating systems such as Solaris and HP/UX because it was never designed to be so. It was originally designed for an x86 platform, and has only relatively recently emerged as a contender in the mid-range server market. Thus it is to be expected that it is perhaps not quite as scalable as its commercially-available counterparts. I doubt that anyone would seriously care to dispute this.
Reliability, availability, serviceability: I believe the same holds to be true. Linux was originally designed as a home hacker's system, not as a mission-critical server platform. While great strides have been made in this area as Intel and other x86 systems have become bigger and better - and thus thrust the PC into the low-end server arena - there is still a long way to go. Beowulf (not investigated in the report) is to my knowledge the only Linux clustering solution currently available. SMP resource management is still rather limited.
Here the report admits to a lack of hard evidence about system stability, at least in terms of mean time between system failures. To quote the report verbatim, "anecdotal evidence abounds." (I've heard of Linux systems whose uptime has exceeded a year, though in my rather limited experience with Linux I have yet to witness this. I've seen uptimes of about six months with FreeBSD, and a maximum of about three weeks under Windows NT. The HP9000 I'm logged into only has a current uptime of four days, but since it's a development machine that doesn't mean a lot.)
Internet functions: the only gripe here is about availability of good commercial e-commerce applications for Linux - something of which the readership here is only too aware, I hope!
Distributed Enterprise Services: likewise.
System Management: The report has quite a few good things to say about linuxconf, and does list a number of shortcomings in the system management tools of alternative OSes, so I don't see a problem here.
PC Client Support: Samba gets a mention (which is good), but again the gripe here is lack of good commercial software for Linux.
So, in short, the report doesn't really slag off Linux all that badly. The general tone of the report, if you ask me, seems to be that Linux is getting there but for any serious large-scale server applications, it is an idea whose time has not yet come. And I for one am inclined to agree. It doesn't yet have the scalability and resilience of those operating systems designed for high-end servers (notwithstanding the operating system itself may be more stable -- but what good is a stable OS if your data is lost on the day when it _does_ crash?) and it doesn't yet have the commercial software to make a good system a great system.
So, in short, I don't believe the report is saying that Linux is a bad system. It's also not telling us anything new. It _is_ a viable system, for the low-end server market; and it's a damn sight cheaper than anything else out there. The WSJ editorialized it to death. The report itself is quite reasonable.
"If it's a bad idea, trash it. If it's a good idea, steal it and release the source code."
--
"This isn't the post you're looking for. Move along."
The jist I was getting was that a Linux box, specifically a Linux/X86 machine, can't scale with the big boys, where big boys are the 64-processor jobbies, likely liquid-cooled.
My three reactions:
1) Linux wasn't particularly built for that. I see no shame in that it can't handle business applications like a monster server. I've never seen a monster server. How many applications are there for a 64-processor server? the only ones I see are monster scientific and graphical work, like ray-tracing and fluid flow dynamics. Such applications work well on clusters of Linux machines. IBM bought a $150K horde of machine last month, installed Linux on the whole bunch, and proceeded to go stride-for-stride on a ray-tracing problem with a $9M Cray YMP.
2) There are good odds that a lot of the limitations are those of the x86 architecture. It's almost a miracle that x86 scales to real servers, considering its bitty-box heritage. If you want big box performance, you gotta buy a big box.
3) Are they trying to tell us that NT is more scalable than Linux? By what criteria? This is the shocking part. I'd say "laughably", but they're serious. Wrong, but serious.
--The basis of all love is respect
I wrote up a nice little response, and had it bounced back before I saw this comment....Anyone found a valid address yet?
--tony
{|}--\/\/\--Tony Hagale -- tony@hagale.net
Even better- on the opening day of a NeXTWorld EXPO (I think in 6/93) the Journal ran an article by G. Pascal Zachary. The writer said that it was doubtful NeXT could ever make any alliances with others in the industry, and thus would surely fail pronto.
That morning, at the keynote, Jobs announced they were working with DEC and HP, and some other things. It resulted in NeXTSTEP running on HP workstations, and NeXT's Distributed Objects software running on HP/UX and Digital's Unix. (The idea being that you could use DO to offload processing to huge big-iron compute/database servers. There was a demo Mandelbrot app which did this.)
In the case of the Linux article, I think they just relied too heavily on the one lame source, and didn't bother checking the facts in the report.
Overall, the Journal does fairly well, as long as you stay away from the loons and fascists on the OpEd pages.
Personally, I'm pissed that they're running ZiffDavis stories. I'm *paying* to read the Journal, not ZD. I can read their crap for free.
The ambitions are: wake up, breathe, keep breathing.
Not many people have played with NT's auditing capabilities much. They're somewhat primitive (tending as auditing does to generate huge logs needing sophisticated analysis to be useful) but when used selectively can be very powerful. Every Windows NT object that has permissions in NT-speak (an ACL) also has a so-called SACL, or security (read auditing) Access Control List. Play around with User Manager on NT (musrmgr for local security accounts, usrmgr for domain) and enable auditing via Policy menu Audit option. If you enable file and object auditing you will find that the permissions dialog now includes options to control auditing per file or directory (which is accomplished by editing the SACL).
This is not a trivial add-on, it's integral to the base security services of the OS. Comparable auditing capabilities are not available for Linux or FBSD to my knowledge.
NTFS uses transaction logs for all filesystem metadata changes, for fast restarts without a full fsck, basically. It uses transaction processing to roll back and/or roll forward all in-progress changes to the disk structure (but not the user's data) and recover a consistent state after catastrophic failure (power or bluescreen) without spending exponential amounts of time based on volume size cross-checking everything.
For more information, see the section titled Recoverable File System here.
Still FUD - do you believe that NT "as available on 1/1/99" is significantly better than Linux? Yet this is what the chart shows...
Regardless of the report the WSJ really took things out of context. And that is where the FUD is coming from. You have to be able to smell it.....
root@localbrain root>ps ax |grep thoughtd
Linux is currently being ported to the S/390.
There was a post on linux-kernel about a week
ago -- perhaps someone has the URL.
If an S/390 CPU screws up inside, the fault
will be detected. All operations in the CPU
are done twice, in parallel, with the results
compared. Any mismatch is bad. Your program
won't even die or get corrupt. It can be moved
to another CPU while the broken one is replaced.
The gall they have at publishing that. MSNBC was a mistake in the first place. The Microsoft Propaganda Commitee has found its portal for FUD.
Madhatter --It's no wonderland out there.
If MSNBC is anything like Hotmail, then MSNBC is running on Unix. Apparantly they've use NT for their MSNBC site, cuz they've been slash-dotted! Right in the mouth of the evil-one.
Madhatter --It's no wonderland out there.
Interesting point indeed. Microsoft has always kept the unix thing about Hotmail hush hush.
Madhatter --It's no wonderland out there.
Good point on this. I meant it mainly as a joke because of the fact its "MS"nbc. What I said earlier was biased in it's own right because I don't have any proof that Microsuck meddles in the business of their shafting partner NBC. All I can do is rant and rave and say "Bill Gates Sucks". I don't want to talk about the good things of Windoze because there is enough mindless sheep out there singing its praises allready.
How's that for biased? Maybe one day we can all get along ?:)
Madhatter --It's no wonderland out there.
Well, aside from the fact that this is an MSNBC article, these guys said that AIX was better than SUN? Bah! They must be high.
I tried emailing them on the link provded. It's a fake email address im guessing. Bounced on me.
this space for rent
>Slashdot has NEVER been a source of FUD against Microsoft
Bull. I've seen more lies about Microsoft propagated on slashdot than I've ever seen lies about Linux propagated by Microsoft. Sometimes it seems slashdot is nothing _but_ FUD about all alternatives to Linux.
Slashdot - News for Herds. Stuff that Splatters.
>Even if they were talking about journaling, NT doesn't do that either
Journaling is an integral part of NTFS. More FUD.
>Linux DOES do SMP
Yeah, so does NT. Neither does it well, though Linux does it better than NT does.
Slashdot - News for Herds. Stuff that Splatters.
More FUD.
>AIX: Logging Sucks.
>Linux: Logging Rules.
>NT: NT does logging?
I'm not sure exactly what kind of logging you're talking about, but I can't think of any definition by which AIX does less/worse than Linux.
Slashdot - News for Herds. Stuff that Splatters.
>I, like others, suspect that when they said `logging,' they meant a journalling file system. Why don't you do a technical comparision of JFS and ext2fs and tell me what conclusion you come to now?
While you're doing that evaluation, don't forget that ext2fs relies on asynchronous metadata updates to get performance, meaning that if you crash your metadata is not guaranteed even to be consistent (let alone reasonably current). Furthermore, the penalty for enabling synchronous metadata updates is - according to the _authors_ - "undetected data corruption". Even worse.
Slashdot - News for Herds. Stuff that Splatters.
Looks a hell of a lot like what I got used to seeing on AIX. I had to look back at the start of the article to make sure that's not what it was.
Slashdot - News for Herds. Stuff that Splatters.
>But I'd like to see next year's report, when they analyze Linux with a distro using the 2.2.x kernel. Then we'll see.
Don't get your hopes up. "Less bad" is the phrase I'd use.
Slashdot - News for Herds. Stuff that Splatters.
>The Journaled File System (JFS) was created by IBM.
"JFS" is an IBM thing, and may have been the first commercially available journaling filesystem (certainly for UNIX) but there were others working on essentially the same thing at the same time or perhaps even a little before.
One thing that people seem to be confused about here is journaling vs. log-structured filesystems. A journaling filesystem only logs metadata ("data about data") such as file sizes and allocation maps. The actual file contents are "someone else's problem" just as in a traditional fsck-style filesystem. A log-based filesystem is much weirder. Everything including file data is logged, and once you have a complete log, well, why do you need anything else? So the data exists _only_ in the log, and their a pruner process used to get rid of old log data that has since been overwritten so that space can be reused, etc. It sounds like it would never work, I know, but it can actually work very well depending on your read/write mix.
Slashdot - News for Herds. Stuff that Splatters.
>Alternatively, one could write a ufs/ffs for Linx
UFS/FFS is not a journaling filesystem, and does not solve the problems that journaling was invented to solve (primarily that of recovery times proportional to volume size vs. proportional to I/O rate).
Slashdot - News for Herds. Stuff that Splatters.
>You can't get highly-available failover clusters with Linux. [though the linux-HA project is working on it]
Progress on this has been glacially so. That may be because the primary author was scared off when I, as one of the authors of the commercial product he was blatantly and transparently ripping off (he makes his living as a support engineer for that product, and had taken classes I had taught), sent him some email a while ago.
Or maybe it's because HA is a hard problem, and there seems to be a much greater interest in the Linux community in dicking around with themes and other brain candy than in solving hard problems.
Slashdot - News for Herds. Stuff that Splatters.
>Congratulations! You have just assured that my company will never purchase any of your so-called "research".
>
>Systems Designer
>TekConnect Corporation
I'm sure that scared them.
Jeff Darcy
Founder, Systems Designer, and Chief Bottlewasher
SuperWhizBangTek.com
Slashdot - News for Herds. Stuff that Splatters.
>I hope someday you realize the futility of silence, Jeff.
Oh, I know the futility of silence. What makes you assume I'm silent about issues that need to be addressed? This particular instance is a poor example of how I deal with such cases, since the only "wrong" thing D.H.Brown has done is provoke the knee-jerk reactions of a bunch of lightweights. When I care, I act.
My intent, beyond a little light-hearted fun, was to point out that if you want to be heard you'll have to do something better than identify yourself as a "Systems Designer" at some company nobody ever heard of. Why should they not assume that TekConnect is just another "vanity company" that only really exists at the local registrar's office? Why should they do anything but laugh at your threat never to buy any of their research when the probability was practically nil anyway? Don't explain the answers to me; explain to them.
Lacking anything more convincing than a made-up job title, it's hard to see your post as being truly aimed at getting results from D.H.Brown. Seems a lot more like dressing up in corporate-big-shot drag for the benefit of the kiddies here who don't realize how easy it is to fake these sorts of things.
Slashdot - News for Herds. Stuff that Splatters.
I guess you havn't played around with any servers then? Many of the large servers use RAID with special SCA connectors. These servers are designed such that you can easily remove a hard drive and replace it while the server is running.
I have done this on numerous occasions. It was pretty cool watching the RAID system rebuild a drive after replacing it without having any down time.
I think IDE RAID environments are now available, and if not they soon will be.
This post is encrypted twice with ROT-13. Documenting or attempting to crack this encryption is illegal.
That looks awfully like VMS error/system logs to me. So, VMS lives on, in a fashion ...
-- Post No Gravy
Still, you miss my point...
,yes, yes...Linux is good, but it's no good for *enterprise* operations...don't stop buying multi-million dollar AIX/Solaris/HP-UX/what-have-you systems...never mind that for the same price as that single 32-processor SuperDuperBox you could buy several Linux machines, along with the people to program your applications and manage the system..."
DHBA is selling a report for a thousand dollars that draws unsupported results from dubious methods. If this is the type of work DHBA and Mr. Iams put out, why should I care to have any knowledge of their past efforts?
I never claimed that Linux is the "be-all/end-all" of operating systems. What I take umbrage at is DHBA's classification of Linux as not being "enterprise-ready". The probablitity that one would be ill-advised to run an "HMO data center" on Linux when compared to other systems is irrelevant to the majority of businesses in the world, who don't generally find themselves running "HMO data centers".
I wouldn't advise anyone to deploy any UNIX variant or derivative as an enterprise-wide end-user office productivity solution (yet); does that make those self-same OS's non- "enterprise-class"? It depends on one's point of view.
I didn't criticize DHBA for using what you term "industry recognized" or "industry standard" benchmarks, I criticize them for NOT using them! DHBA attempts to support much of their claims concerning "scalability" (another term with many, varied definitions) with the assertion that "no published figures are available". How preposterous! If you can't find figures, build some boxes and test--that's *real* research.
What is to be gained from this? How about truth, facts, and scientifically supportable conclusions?
Why do I think that this firm is defending the status quo (by which I mean established UNIX-variant vendors) at the expense of Linux? Because they are using the current hot topic to generate sales of a poorly-researched report by playing upon the fears of most IS managers that they might be doing their jobs in a better fashion , i.e. not wasting so much money.
"Oh
I can see that you *do* have a reason to protect your identity. If Slashdot readers knew who you were, they would be advising others to steer clear of you and your company, if reading reports like this pathetic DHBA trash is how you draw *your* conclusions.
BTW, judging from the way HMO's conduct their businesses, I'd say they have much more to worry about than Linux vs. other *NIX'es...
-m
I hope someday you realize the futility of silence, Jeff.
I couldn't really care less what people assume about me or my company. If you want to know more about me, try taking a look around on the Internet (perhaps you've heard of it?). Whether or not you or anyone else has heard of me, the company, or anything I or we have done has no impact on my self-esteem.
The point I was trying to make is that rather than make glib, purile comments like the one you made, perhaps you should consider making your opinion of the business and research practices of companies like DHBA known. If everyone sits back and listens to this kind of crap without responding, we are all doomed to be deluged with more of the same.
I didn't assume anything about your personality; I responded to the statement you made, which had absolutely nothing to do with the issue at hand. Like so many other posts on Slashdot and elsewhere on the Internet, it accomplished nothing but to waste bandwidth.
Perhaps if you resorted to more mature methods of communication than attacks of a personal nature, people would take you and your opinions more seriously. As it stands, all you have done is to prove your own point that this issue has provoke knee-jerk reactions.
My intent in responding directly to DHBA was not to "scare" them, as you seem to believe, but to inform them that they have lost another potential customer by publicizing the fact that their research methods are lacking in integrity. Frankly, I'm rather tired of that kind of crap, and I felt like saying something about it.
What I didn't realize is that I would be subjected to simple-minded comments like yours, which mistake my lack of willingness to spend time instructing those whose integrity I question for a lack of experience, knowledge, or ability.
For the record, I would certainly not consider myself a "corporate big shot", and I certainly don't need to fake anything. I simply provide my identity as a matter of course when I correspond with others, as any *adult* should. My identity is easy enough to verify, should any desire to spend the time doing so.
More to the point, why should DHBA assume anythng other than that I am a potential customer? They may laugh all they like, but they will still not receive any money from me.
Where is it written that only famous people have opinions that matter? Only an utter fool would entirely ignore the voices of potential customers. Every voice is important in some way; a million whispers at once make a very loud sound. I have added my voice--where, sir, is yours?
Unfortunately for us all, you seem to think your voice is more important than most. Equally unfortunately, you seem to care quite a bit less than you act, and act irresponsibly when you do care.
If you're really interested in sending e-mail,
write directly to D.H.Brown, like i did
Their address for the report is the following and
there's a link for feedback on the site!!
http://www.dhbrown.com/dhbrown/linux.html
Go easy on them, you can't read the full report so some things might have been exagerated by WSJ!
"Code free or die!"
If you want to know _exactly_ what the computer is doing... Look at the source.
The source doesn't tell you what it is doing. All the source can tell you is what it will do when it is in a specific state and gets a certain input.
--- Abigail
is it time to battle FUD with FUD?
As far as journalling file systems go, neither SCO ODT nor NT have it. Ok, the newer Unixware has it, and NT 5 *will* have it; but NT in its current forms doesn't.
As far as SMP goes, the company I work for recently took down an SMP NT RIP (2 processors) and replaced it with a uniprocessor machine because the technician told us that performancewise adding a processor to NT does *nothing*.