Slashdot Mirror
PCMag's PCTech Reviews Linux Kernel 2.2
Gryphon writes "PC Magazine has published a pretty level-headed 8-page review of the Linux 2.2 kernel. Mostly a features review, compared to Windows NT. I think this is pretty significant, considering a lot of Windows users (including me in years now past) read that magazine! "
It's pretty standard ZD fare, in that it's light'n'fluffy and stays away from anything really technical or contraversial, but it's also well balanced and very fair.
:)
More suprising was the word "Amiga" in a ZD article! Wow!
It would be interesting to see a more in depth, technical comparison between say, NT, Linux, AIX, FreeBSD and Solaris running similar programs on similar hardware. I'd really like to see where Linux does fall down technically - so that we can fix it.
They say there is no need to recompile because you can reconfigure your kernel via its virtual file system ????????? what the hell are they talking about ! modules perhaps ?????
I just hate it when these magasines just throw some random tech-speak, only showing off that they do not know what they are talking about...
This says it all. Mediocre writer and editor.
Now for some additional facts:
1. NT is _more_ difficult to install than Linux
2. NT need for reboots is not `good'. It's bloody
BAD.
3. PnP is a Win9x thing. How come then the table shows PnP `Good' for NT?
4. The current SMP sweet spot is 2 processors for Linux.
5. NT is _NOT_ more user friendly. Unix/X is.
See, if this was just a hobby for ya anymore, sleeping in would be cool... but, since you're "Blockstackers" now, maybe it's time to set the alarm clock, hrm ;-)?
I didn't realise that capabilities were introduced with linux 2.2.x :-)
I know that this is not what the author means in his introduction, but he should be more careful about using words which have specific technical meaning in the context which it appears, but intending the "everyday" meaning.
All windows machines can have IP masq through WinGate. AFAIK a lot of people use it with RoadRunner. But that is just a program, not a kernel feature, so Linux still has the edge on that. Plus WinGate supports IP tunneling of other protocols only due to the screwed up IP implementation in Windows.
Your point of different windows examples that they used is valid though. Windows 9x supports a lot of hardware, but there is much more to that. Linux has most trouble win winmodems, as does NT. Only win9x can make use of my piece of shit of a modem that came with my hp pavilion (again as far as i have tried). other than that linux does just about as much as NT if we don't consider things like 3D accelerated drivers for some video cards (though most 2D chipsets are supported) and some SB cards.
if youre gonna run linux on a 386 to do anything
similar to what youd do on a pentium/166 w linux or NT
its not gonna work. first of all youd need a shitload
of ram because linux swapping on a 386 is slower than crap, esp with the shitty ass
hard drive subsystems that 386s tend to have. so after you cannibalize a bunch
of other 386s for ram, your gonna have a bunch of fat bloated apps that have been optimized for pentiums unless you compile stuf yourself.. oh bo compiling on a 386
thats fun. ok you could set up a cross compiler.. now... with all the time youve spent it would have been faster to just buy a damn pentium 90.
i cant think of anything a 386 linux box can do well except maybe be a firewall
or a print server for some piece of shit line printer.
I ordered the 6.0 Core package from Red Hat because I wanted exactly one thing out of the 2.2.x kernel series that the old linuxhq page said it would have. I quote from the new one (which has the same documents): "Performance of Linux 2.2 with 16Mb or more RAM is noticeably superior."
I have 96 Meg, which fortuitously 2.0.36 recognized immediately, and I would love to see some more performance.
However, I still don't know where that increased performance is going to come from. What are the specific benchmarks?
Thanks in advance.
Mitch
Hrm, since when did "level-headed" become a synonym for "loaded with hype"?
I totally agree...
From Partition Magic Manual ('Basic Concepts' page 6)
'Fat32 is a file system that can be used by Windows 95 OEM Service Release 2 (version 4.00.950B), Windows 98, and Windows NT 5.0. However, DOS, Windows 3.x, WindowsNT 3.51/4.0, and earlier versions of Windows 95 cannot recognise FAT32, and are thus unable to boot from or use files on a FAT32 disk or partition.'
Maybe they are considering NT 5.0.
Moreover, I thought Linux could read Mac partitions (I don't remember the exact name).
I think they compare to Win200, but then mistakenly put NT on the chart.
So if the chart included "Is a shipping product"
M$ would have gotten a big fat NO.
In the side by side, he states that NT 4.0 has usb
support and linux doesn't. Not at all true. I'm using my usb mouse with Linux today. If it's about distros, there are distros that ship with USB (for iMac).
In anycase, no one mentions how microsoft treats users of it's older operating systems. Try to find a page with the patches for win95. Frightening...
Good points. But the chart (which will be used as the summary) lists MSWinNT, not MSWin2000.
MSWin2000 is *NOT* a shipping product, Linux 2.2 is. And when (no, if) MSWin2000 becomes a shipping product then at that time Linux will better yet.
The reason they didn't compare against 2.4 is quite simple: at the time they wrote the article, 2.4 hadn't been released. Magazines like PC Magazine have a *3 month* lead time. Software companies like Microsoft know this, so they give reviewers advance copies of software they know will be released by the time the magazine hits the shelves.
check out linux/drivers/block/md.c, and search for 'hot-add' - Linux clearly supports hot-add. Plus, the Red Hat 6.0 'raidtools' RPM includes two new utilities: 'raidhotadd' and 'raidhotremove'. The RAID-HOWTO states that RAID arrays can be started up automatically at bootup - even as a root filesystem. Maybe the author has mixed it up with the 2.0 version of RAID?
Yes, my Alcatel ADSL "modem" is actually a bridge from ethernet to ATM - i.e. it's transparent to ethernet. The booklet with it doesn't mention any remote admin of the bridge, so it doesn't appear that any "support" is needed, especially in the kernel.
(If the kernel has ATM support, I can plug into the box's ATM port instead of ethernet, but no biggie.)
Hopefully no one will be dissuaded by this error in the article. Broadband is very important to people, so we need to make it clear that ADSL and cable are no-brainers.
I've said this some time before:
Linux will kill NT in Y2K!!!
I'm close to the reality now...
Fortunately the third party product you are talking about works really well and appears to be one of the more stable bits of my NT system. ISTR it wasn't that expensive either.
high hopes d00d.
In terms of features, Wingate is about the same as IP Masquerading (though I'd like to see wingate handle a cable modem on a slow 486). It does require some client-side hacks, however those aren't very difficult. There are some proxy servers which behave exactly like IP Masquing (WinRoute comes to mind), however those aren't the norm. The price of course is the issue. Most people wouldn't want to pay that much, however most people also wouldn't want to learn how to use ipfwadm or ipchains. Then again, most people don't even know this stuff exists, and therefore it doesn't really matter.
However Windows 98B and Windows 2000 will both come with IP Masquing built in. They don't call it IP Masquing, they call it something else which I have forgotten, but it's the same idea. I haven't heard much about how well it works, however I'm sure I'm going to hear people pretty soon wondering where they would be if Microsoft hadn't invented that great technology.
NT is a funny beast. There is an NT box in my office that has been running for months, flawlessly. It is an NT server box, but it never does any serving, just Word, Excel, and few exotic geophysical applications. It never gets turned off, and runs very sweetly. In contrast, my NT box doesn't exactly crash, but the explorer applet segfaults on a regular basis. This is more an annoyance than anything else, but it reminds of gnome! At home, the NT side BSOD's about once a month on boot up. The linux doesn't kernel panic, but sometimes X locks up so bad the keyboard and mouse quit.
Strange stuff.
make xconfig anyone? yeah, make isn't a button, but still....
/boot/vmlinuz
Oh, and neither is
make dep; make zImage; make modules; make modules_install
cp arch/i386/boot/zImage
lilo
/sbin/reboot
:)
ZDNET in commenting on GUIs for Linux 2.2 amusingly introduces us to the "Full-Moon Effect" whereby "a graphical operating environment sprouts the fur and fangs of a command-line window whenever unusual conditions arise." As the GNOME graphical environment and KDE mature, the hairy spectacle will become a rarity. See here for the full article.
Claims that the memory management in 2.2.x kernels is much better than in 2.0.x are rarely backed up by figures. Can someone supply details of specific benchmarks comparing performance between 2.2.x and 2.0.36?
PS. 2.0.36 with 96MB RAM seems to run excellently in memory-sufficient circumstances, but when individual processes approach about 80MB resident set size, swapping and other MM overheads cause a severe loss of performance.
NT does not allow for card hot-swapping and there is no CardBus support either unlike with D. Hinds PCMCIA package. No, I would not call NT's PCMCIA support good.
Linus intended Linux to be a Posix compliant OS
from the very beginning. He was unable to obtain
the POSIX specs, though (linux/README says "it
aims toward POSIX compliance", but I think it goes
further than almost any other OS towards this
compliancy).
Speaking of which, my penis is one node in a beowulf cluster. I'm looking for another node, preferably a vagina. Any takers?
This report seems OK, but some things are
not exactly right.
The author says PCMCIA support in Linux is
poor. I say its SUPERB!!!. The package
of David Hinds is one of the perls of Linux
kernel developement.
Ok you won't find PCMCIA drivers in the
kernel on ftp.kernel.org but they is an
extra seperate package. Which is I think a good
thing anyway because linux-2.2.9.tar.gz is getting
to big already.
Anyways, another Linux succes story in a Microsoft
oriented Magazine. Not bad, not bad at all!!
I have to admit that I use NT4 40% of the time, BeOS another 40% and Linux like %20, and there is a way to get USB under NT4. You need to have SP5, then you have to download the USB support from Win2000 (it's a hack....), and then USB will werk. BUT it only werks with like some video camera, mice, and keyboards, so it's semi-pointless.
Please wake me up, I must be dreaming that my PnP sound and network cards configure themselves...
;-) ]
:-P ) :-) ) I've rebooted it other times, mainly to play with the Hurd, but unless I'm playing with things marked "do not install unless you really really want to risk your stability" it..just..never..dies.
[ On filesystems: The reason it was stressed, I suspect, is simply that Linux supports more. Lots more. Lots, lots more. NT supports, what, FAT16 and NTFS? I believe I found a Win95 program to read ext2 partitions but that hardly qualifies as 'support' -- in particular, I can't mount them as proper filesystems. A quick scan of my kernel config shows 13 possible foreign filesystems, if I compress all the offerings from Microsoft into one entry. And I'm not even counting the network filesystems. AFAIK NT only supports NCP and SMB natively. If you want to include 3rd party drivers you should also include all the 3rd party kernel drivers floating around, which include a number of versioning and journaling filesystems...I think at least one of them compiles
AFAIK the Linux API does not change for 64-bit processors (correct me if I'm wrong); I consider it to be a sad comment that Microsoft even has to publish new interface specifications. (I guess hardcoding variable sizes into their type names caught up with them?) Most well-written code should just need a recompile. (now that I've said that, I'm sure that all my code will fail on a 64-bit arch
I'm rather confused about the occasional assertions that NT is more stable than Linux. 'as stable' I can see. But the only time I had to reboot Linux lately was when the filesystem code freaked because of physical damage to a floppy. (I think this is inexcusable personally, just so you know
Daniel
Gee, that isn't a very wide range. How about "Palm pilots to clustered supercomputers"
----------------- ------------ ---- --- - - - -
----------------- ------------ ---- --- - - - -
Your honor is perfectly understandishable.
Seemed good, but at a quick pass, I noticed it said *NT4* had usb support..... if it did, my friend shawn wouldn't compilain so much. :) No time now to look for other errors in accuracy, but seemed like semi-informative article overall.
This sig left intentionally blank.
An article like that might not seem like much today, 2 years later, but back then that was great press coverage. It's nice to see he's still on the Linux bandwagon.
Oh god... I read an issue recently in which Dvorak's editorial advocated nationalizing Microsoft-- it's time for Windows to become a national commodity, basically. This has to be the worst idea EVER. I can't take the magazine seriously anymore.
Not bad... I only caught a few technical inaccuracies.
NTFS support is read only.
2.2 includes an experimental read/write NTFS driver.
DSL and cable modems are mostly unsupported.
Every DSL or cable modem I've ever seen has been an Ethernet device, and Linux works fine with them. If they're talking about phone/cable company support for Linux, that may be another story, but that's a PHB issue, not a technical problem.
They state that SMB is a networking protocol, like IPX, rather than a network filesystem, like NFS.
They missed a couple of ports and a couple of filesystems, but, hell, I can never remember all of them, either...
And they didn't include Slackware in the list of distributions.
I believe the term you're looking for is "geek".
Heehee... I run RC5 on all of my boxen, too... the 386-16 gets about 5000 keys/s. That's about 1 2^28 block a day...
My Celeron, meanwhile, will rip through a 2^28 block in five minutes...
I personally have Linux running on four different 386es.
There's eddi, a 386SX-25 with 4M of RAM and an 80M disk, which is my laptop. Not a blazing fast machine, but she was cheap ($20), she's portable, and she's sufficient for carrying work around with me. You don't need a lot of space or processing power for writing code, or even for compiling small projects. She's currently running 1.2.13, because she hasn't got enough disk space to compile a 2.0 kernel.
There's deliah, a 386SX-16 with 8M of RAM and no disk. She boots a 2.2.5 kernel off a floppy disk, then configures her ethernet and mounts her filesystem using BOOTP and NFS. I use her as a not-entirely-dumb-but-not-very-smart terminal for my faster machines.
There's gabrielle, a 386DX-40 with 20M of RAM, a 120M disk, and a 1.0G disk. She runs X, and I use her as an X terminal. Mostly I run stuff on other machines with the display redirected, because she hasn't got enough horsepower to handle X and, say, Netscape at the same time. I also use her as my guinea-pig machine. Because she doesn't do anything mission critical, I use her as a test bed for new kernels, new libc installs, and such things. If I screw her up, no big deal... I could wipe the system and reinstall it without losing anything important (I haven't had to do so much as a floppy-rescue yet... which is fortunate; gabi's got no floppy drives). She's currently running 2.3.5, and if I get sufficently bored today, I may download 2.3.6pre1 and compile it. Sure, it takes four hours, but there's no reason I have to sit around and wait for it.
The last one is leviathan, a 386SX-25 with 8M of RAM and a 120M disk that I'm planning on embedding in my dashboard as a CD player and radio as soon as I get the power supply for it built. It's currently up and running in a caseless heap on my card table.
I'm not even going to get started on the 486es I'm running Linux on...
I'm not sure, but I think the crucial difference
is whether it is true one-to-many IP multicasting
or is it pseudo-multicasting: ie. a series of
one-to-one connections transmitting the same data.
This review suggests that NT kernel is currently not capable of doing true IP-multicasting.
I commend this review for at least getting
these things right for once:
1. Comparing Linux kernel features to NT kernel
features instead of doing another "Linux/Apache
vs. NT/IIS" snow job.
2. Pointing out that the Linux installation
process depends on the distribution you use,
and yes, the two remaining daunting areas
for newbies in each installer is disk partioning
and the video card and monitor settings for X.
3. Nice explanantions of why new features such
as IP multicast and frame buffers console are
important.
4. Pointing out the areas where NT will be playing
catch-up with Linux such as supporting Merced.
Aside from a few accuracy flaws, it looks like
ZD finally hired a reviewer who actually has a
clue about "this crazy Linux thing all the kids
are talking about."
I like to see this kind of good press on Linux. Errors seem to exist in there but overall pretty nice, positive article.
Da -941
#941
And compare the press coverage of the 1.0 kernel debut :) I was happy to read a level-headed, non-fudish not-too-technical appreciation of the 2.2 kernel. As pointed out above, there are some errors, but they don't seem intentional. Also, did anyone notice that he refered to the X Window system and not X Windows? That's pretty good.
"shop smart:shop s-mart" ash
AFAIK, NT4 has a PnP service, although it doesn't seem to be any better than Linux's PnP capabilities. Also, NT's UI is rather user friendly for those tasks for which you would rather use Win9x. However, I agree that the same UI makes NT server administration a big ole PITA.
In Soviet Russia, Jesus asks: "What Would You Do?"
Yeah, you're right I forgot to mention PCMCIA, but that's because I didn't go that far into the article. I suppose that the author of the article didn't bother to install the pcmcia_cs package (included with just about any Linux distribution on this side of the galaxy). This makes me believe that the writer was only cut&pasting from other texts and documentation to write the article.
In Soviet Russia, Jesus asks: "What Would You Do?"
The non-technical nature of the article seems to be due to the author's unfamiliarity with the subject. There are many lines that are either pseudo-typos or flat-out misunderstandings. A few examples:
I don't think I'll go beyond the third page...
In Soviet Russia, Jesus asks: "What Would You Do?"
I didnt know about this feature. Where do I find more info about it? Whats the connection to VFS?
There is no such thing as luck. Luck is nothing but an absence of bad luck.
It's amusing to see them talk about filesystems
that arn't really filesystems....
For every problem, there is at least one solution that is simple, neat, and wrong.
Thats a really good point. I didnt even think about my friends, who are now linux users, who started out as Win Quake players.
John Carmack is a remarkable individual who has done much for the linux community (and for cumputing as a whole) by forcing hardware manuf's to adhere to standards, and also by forcing superior standards (OpenGL, MesaGL, etc.) into the limelight.
In retrospect, Quake seems like it could be the perfect teaser to get people interested in Linux.
Four-digit slashdot ID. Recognize.
It seems to me that the average Joe computeruser, even the ones who are really into gaming or whatever, either has not heard the word linux or has no idea what it is. Sure, articles like this one help, as it appeared in a widely-read Windowsesque magazine. But to most, linux is still freak show material.
Some recent improvements have made things better. Though the actual distribution (IMHO) is about as stable as a drunk prom date, the Red Hat 6 installation process was fairly painless (I have always and still do run slack). Personally, I dont care about installation proceedures, but to sway sheep to the flock, it's got to be idiot proof.
Idiot being the key word.
Positive things are happening. I think that with time, the product will improve and 'sell' itself. We simply must be patient, and continue to hone linux into an even more robust and powerful OS.
Four-digit slashdot ID. Recognize.
www.pcmagazine.com tuns apache
Check it!
Four-digit slashdot ID. Recognize.
Am I missing something or does NT4 really include IP masq? Didn't they say it will only be available in win2000?
Also, did anyone else notice that they were comparing Linux to NT4, win98 and (yet unreleased) win2000 at the same time? Win98 supports pnp better and NT4 supports RAID better...
They didn't make clear *which* OS they were talking about. Last I checked NT didn't support USB, couldn't read FAT32, and had no pnp or power management support whatsoever.
___
If you think big enough, you'll never have to do it.
Undoubtedly, they were talking *useful* or *mainstream* ports. I wouldn't call the "port" to the Dragonball CPU on the PalmPilots terribly useful for anything other than "gee-whiz" sort of thing.
I'm not sure where the reviewers found problems with pcmcia card services though. I know that while my laptop that Win95/98/NT all had problems with card services, I was able to get surprisngly good results when I installed linux there.
I suppose hotswapping may be an issue, but when you run with the same cards in there all the time as I do, I don't hit that wall.
--- http://foo.ca
Well, I for one was not aware that NT supported FAT32 at all. I have just built (reluctantly) an NT machine with Service Pack 5 (the latest) and there's no mention of support for FAT32.
It would have been nice if the article mentioned the recent open sourcing of XFS, which eats NTFS and Ext2fs for breakfast. Oh, and I've heard rumours of an upcoming Ext3fs. Any word on its capabilities?
See http://www.winternals.com - they sell the FAT32 driver - the read-only version is free of charge, the r/w version is chargeable but not too expensive. I used it when I was dual booting Linux and NT, now I just use VMware (http://www.vmware.com) to run NT on top of Linux.
Windows 2000 supports FAT32 r/w.
Agreed - I found that Red Hat 5.2 just recognised my PCMCIA modem/network card (from Ositech) on a Thinkpad 755 and the installer even let me NFS-install over the network using this card!
This looks a lot like the press release which was going to be sent out when Linux 2.2 finally came out. If only I could find a copy of that and compare... :^)
This
The minimum requirement for NT 4.0 is something like a 486/33 with 12 MB RAM (for an Intel-based machine) and 120 MB of hard drive space.
For a RISC-based machine, you'll probably need 16 MB of RAM.
On the other hand, I just worked on a P90 with 24 MB of RAM today, running NT, and it was unusable.
Some minimum requirements are more reasonable than others.
--
QDMerge -- generate documents automatically.
how to invest, a novice's guide
Level-headed, for sure, but it's nothing new for most of us.
;)
Decent article: my only complaint is that they don't talk about the emerging (!) GUI interfaces for Linux very much at all, solely relying on the "users have more control over their system" argument -- an apples to apples comparision would have been a little more handy.
Good article, though, for trying to convince familiy to switch over
I would have liked it better if the author didn't have to throw in that snipe about having to go through a repartitioning. Users are only forced to go through that process since MS had pretty much forced OEMs to preallocate the entire disk to Win9x before the computer leaves their manufacturing facility.
Give the customers (remember them? They're the only ones who are always right!) the option to either:
It would be nice to have a fifth and sixth option:
but which Linux distribution would the OEM select? [Debian|Slackware|SuSe|Caldera|etc.] bigots would surely take to task any OEM that decided to ship Red Hat.
[DR-]DOS?! Are you mad?! Not really; just practical. I still find DOS useful for running diagnostics and/or setup utilities that manufacturers ship with their cards. Until the world standardizes on each machine having a bootable CD-ROM drive and all cards come with their utilities on a bootable CD, I think we're stuck with needing something that can launch our diagnostic/setup software. Since a lot of these utilities would be difficult or impossible to run while Linux was running, it doesn't bother me too much to keep a bootable DOS floppy or a 5MB DOS partition on the first hard drive. (I'm confident that the need for these will go away someday soon!)
CUR ALLOC 20195.....5804M
That's because WebBench is a useless benchmark. Anyone can dump static pages from memory to an ethernet controller. WebBench is little more than an ad for IIS. Does it matter how fast a web server is if my clients are at the other end of a 128kbps pipe or worse?
Apache isn't designed to be fast in these circumstances. It's designed to be stable and flexible. There are other web server choices out there that would be faster in WebBench, including probably 10 lines of Perl code.
pooptruck
I was happy to see "Open Source Code" at the TOP of the comparison chart. While not the most technically literate article I've seen, it shows signs that they're starting to "get it".
-- Don't Tase me, bro!
Ok, quick question...
WHEN did Linux become a POSIX compliant Unix?
(As opposed to a unix-like system)
"That revolution is Linux, the POSIX-compliant
Unix operating system, now out in Version 2.2
--a significant new update."
First paragraph...
From what I gathered from the article they are more complaining about the web servers on linux then about linux itself
The comparison chart impressed some of the more "hardcore NT" users where I work. I guess they never thought something could support more hardware (processors) and have more features than NT :)
Even though it's nothing groundbreaking, it's still positive, mainstream press, and that is what we need.
2.4 been released by now, either. Nor has Win2k. What you say is likely true, but my point was that the article compared a released product with an unreleased product. The least they could have done was compared vaporware with vaporware and talked about 2.4's journaled file system and enhanced scheduling. (no, I don't know those things for certain, those are just two current hot topics).
Your right to not believe: Americans United for Separation of Church and
It's disappointing to see the current proven functionality of Linux 2.2 compared against Windows 2000 in this article. I'm sure many admins are deciding whether to upgrade to Win2K or to Linux, and they would like to see this sort of comparison. But if we're going to compare a product which has been released and thoroughly used for months with a product which may not be available for months (unless you're a beta site) then we might as well compare against a future Linux release as well.
PCmag would have done better to compare the current Linux 2.2 kernel with the current NT4 release and Service Packs available in stores for the bulk of the article. Then at the end they could state that Win2K promises these additional features, and Linux 2.4 will have this other list of additional features over 2.2. That would be a little more honest of a comparison.
Your right to not believe: Americans United for Separation of Church and
I think NT has USB support through one of the service packs; there was an SGI NT machine at school with a USB keyboard.
However, Windows 98 cannot use USB devices, even keyboards until it loads the USB drivers. that was a nasty surprise when installing Win 98 one day; especially since the BIOS could find the keyboard.
Chalk it up to USB being still fairly new, I suppose.
Remember the "unbiased" comparison table somebody put out several months ago, to 'help managers make an informed decision?' It was a joke. After going a quarter of the way through it, it was obvious to a reader with an IQ 60 that the author was slamming NT every chance he got. Nobody would ever take it seriously as an unbiased comparison.
This one is refreshing. It has yesses and nos on both sides, and gives both systems credit where credit is due. I think this comparison will be taken seriously by a great many people. Good work PC mag!
Vidi, Vici, Veni
I would agree about PCMCIA support not being so bad under Linux. I have a MegaHertz dual function (33.6 modem, 10mbit Ethernet) PCMCIA card that I use in a ThinkPad 355Cs. SuSE found it for both modem and Ethernet during install with no problems, and this was with a 2.0.x kernel. I've also had occasion to plug in a few other PCMCIA cards (such as a 10/100 LinkSys Ethernet card) and they were automatically found and the proper modules loaded...
Seems like everything he writes or "predicts" is hogwash. No matter how inevitable something is, if John Dvorak writes about it, it will fail. ;-)
cpeterso
And then there's my 386/40 that's been running the RC5/DES client without a hiccup for the last 423 days and counting. 20,000 keys/sec is nothing to snicker at! :-) You can read about this machine, called Nodens, and some of my other machines, here.
make xconfig anyone? yeah, make isn't a button, but still....
>All windows machines can have IP masq through WinGate.
You mean Sygate? Wingate is a proxy server (if that's what you're talking about.) Sygate lets you configure your client machines to a default 'gateway' without individual apps being 'proxy-aware'.
And if so, a 3rd-party add-on package, even at $80 shareware, is hardly support (as you point out). I think its fair to compare features that are in the Linux *kernel* with things that are in the Win32 *kernel*.
If you are talking about something else, I apologize. I wasn't aware of anything available in Windows to do masquerading. Been using Linux to do that for a while since I gave up on Windows cause it sucked and would'nt do what I wanted (run a decent firewall on my 486 to gateway the Linux and Win PC's in my house to my cable modem.) Oh, and port-forwarding to my quake server, can Windows do that too? Sygate couldn't handle any kind of inbound services at all.
Who else has noticed that slashdot is listed at the end of the article as a `recommended resource`? People are going to come here to learn more after reading that article. We ought to give them a good impression.
To everyone in here that is ripping the minor points of this article, calm down!
In comparison to past articles, this one shines as being quite fair. Remember that PC Mag might as well as been MS Mag a year ago. I don't think a Linux magazine would have been so friendly to Windows 2000. Further, PC Mag was NEVER as friendly to OS/2 as this article is to Linux.
I think it is OK to compare Linux 2.2 to Windows 2000. NT4 has been around for a while and most people looking to put a server into place over the next few months will be looking at Windows 2000 and Linux 2.2. 2.4 is very far away and should be not be considered.
It didn't bother me that the auther switched around between Windows versions. The article is not about which is a better server, etc. but just a general feature comparison. If you are in the Windows world, you get this, in the Linux world you get this. Most admins are not concerned about getting PnP on servers, but are very concerned about getting it on workstations.
The author did refer to features that are not available yet, but will be in Windows2000, but he did the same for Linux. For example, he said that IP tunneling on Linux only does IP, but also said that other protocols will not be far behind.
The author also didn't say Linux was lacking a feature simply because there wasn't a button to activate the feature. Never have I seen the flexibility of the sysctl stuff in /proc discussed in an article. This guy did even though there wasn't a point and click interface.
He even pointed out that even though Windows may be more user friendly, Linux users love the control they have with Linux!
In future, please calm down and treat an article that is decent as it should be treated. Just because it is in PC Mag doesn't mean it should be ripped to shreds. Otherwise, the media sees the Linux community as a bunch of religious zealots and not the serious group of users that just want a good OS that we are.
May the flaming begin... sorry Rob.
It lists PCMCIA support as "poor" - I don't know where the writer got this tidbit, but I've never had any problems with PCMCIA under Linux (with a variety of different notebooks)
Something that astounded everyone here in our tech department was the fact that Linux recognized and installed an IBM Home&Away modem/NIC with no problems at all... this after they had given up trying to get these cards to work at all in Win98/NT.. (they could sometimes get the NIC part to work under 95, if they screwed around enough, using exactly the right combination of DLLs & drivers..)
Just my 2 bits..
The comparison is NT4 vs Linux 2.0.3something...that means, it's just as worthless as all the previous ones :)
:)
I'll belive those numbers when it's NT4 with latest SP vs. Linux' latest stable kernel...till then, those tests are just Microsoft ads
Vox
Pain is the gift of the gods, and I'm the one they chose as their messanger...
Actually, in the same issue they do their own little messed up comparison of web servers, which I tried to post also three weeks ago in the aftermath of Mindcraft. Well, go there and read it now.
The ability to run a great OS like Linux on really
old hardware (which 386's are, in computer time)
means that poor countries in Africa and Latin
America can get a 'net infrastructure up and
running with less effort than by using NT.
The fact is, MS and hardware vendors want you to
buy new hardware -- it's no surprise NT needs big
machines to run on. Everybody "wins" -- even Linux
users, as old-but-perfectly-good machines flood
the market, keeping computers affordable for more
people.
/g.
Ask not what GNU can do for you, but what you can do for GNU.
I looked at the article and, as much as I love linux, I thought the article over-hyped linux and mostly compared NT and linux on points those of us in the linux community like to brag about. There was some negatives said about linux, but if doing a fully comprehensive review, it would need to address more issues that would concern average windows users. Maybe I missed something in the article, but I did not see much of a discussion of user-interface friendliness. NT is a bit more user-friendly and has more unified appearance. X offers moore flexibility with less user-friendliness (though even brain dead people should be able to figure out how to work a window manager, even if they don't look EXACTLY the same.
Just my two cents..
XML is like violence. If it doesn't solve the problem, use more.
With the rash of 'pro-NT' articles and benchmarks out lately and the uproar on Slashdot about the bias shown, it is disappointing to see that even an article like this which is fairly 'pro-Linux' is still criticised when it paints the 2.2 kernel in a better light than it really should be, and includes some blatant FUD about the state of the NT kernel.
Looking at more specific parts of the article:
"You also need far less hardware to run Linux than Win NT; a good old Pentium/166 is just fine, and you can even press that 386 doorstop into service." I will agree that a 386 won't run NT (it requires the CMPXCHG instruction on the CPU), but the inference that NT doesn't run on a P/166 is just ludicrous. I happily run NT Server on my 486/33 at home to serve 98, NT and Macs just fine. There really isn't an issue with performance at all when using EISA SCSI controllers.
NT is cast in a bad light for running Apache slowly, but no mention is made of IIS which actually won the 'Editors Choice' award for the best web platform in the very same issue. Again, showing considerable bias to Linux.
Now for the Kernel FUD:
"It's unlikely that Windows NT will offer 64-bit support until significantly after Merced's introduction." Windows 2000 already has a 64 bit version (Win2000 Datacentre) in development which runs happily on the Alphas. I agree that Linux is already 64 bit on the Alphas, just the inaccuracy of the article's comment is unbelieveable. The Win64 API has been published for at least 12 months now and the SDK is widely available for making apps that behave on 32 and 64 bit platforms.
Mention is made of the improved SMP support in the 2.2 kernel, but it is cast as 'better' than the support that has always been in the NT kernel. The fact is that the 1993 NT kernel offered better SMP support than the current Linux kernel by offering fully reentrant APIs. The 2.2 kernel still has a long way to go before it will scale as well as NT, just as NT still has a way to go before it scales as well as Solaris. The inference that the 2.2 kernel scales better than NT is just FUD!!
The lack of PnP support is quickly glossed over. Again, hiding the weaknesses of Linux and the strengths of Win2000. Even NT4 had limited PnP support!
2.2 is praised for not having to recompile to recofigure the kernel. Isn't that exactly what NT has always done? While recompilation is an added bonus for Linux, surely the praise for not having to recompile must also be thrown back to NT. More bias!!
3rd party file system support is highlighted for Linux (NTFS etc.), but ignored for NT (FAT32, ext2, HFS etc). How is this comparison even close to fair?
NT could always boot from a FAT partition, so UMSDOS is hardly 'non NT'. SMB and NCPFS are not 'beyond NT'. NFS is readily available for NT, and I'm not sure about the availability of then others. Still bias and FUD here.
Linux is consistently written up as soon to support something that NT already supports (eg IPX over VPN), but NT lacking support such as IPv6 is hardly written in the same light. NT does not support IPv6, but is this an issue in the short term? NAT seems to be alleviating most of the problems with IPv4 for now and the NT stack will be available in plenty of time.
Linux IP configuration and route configuration is praised, but no mention is made of any of NT's advancements from NT4 to Win2k over Linux. More FUD and bias.
On page 5 we get a whole three paragraphs (after 4 pages of FUD) stating the shortcomings of Linux. Wow. What a break!! Come on - the match isn't that close yet.
To recap on the FUD in the table on page 6:
64 bit *is* supported to a small extent on NT4EE and is supported on Win2k Datacenter.
The File Systems are complete FUD as it does not include 3rd party support in NT. UMSDOS is a red herring as NT supports FAT booting natively.
SMP support is much better in NT with reentrant syscalls for I/O.
NT's RAID support is much better.
I haven't mentioned that I've found NT more stable than Linux in my home environment (less reboots), but that's a personal thing that mainly stems from the fact I know more about configuring NT than Linux. We have NT boxes at my work that go years without reboots, but empirical evidence seems to be contrary to my experiences.
Basically this article is almost pure pro-Linux FUD. Some concessions are thrown to NT to make it *seem* fair, but it simply isn't.
Perhaps if the Slashdot community values their integrity, they will also write to PC Mag and demand fairer reviews from both sides of the fence? Somehow I doubt it though. Most Linux advocates lose their integrity when forced to admit NT is sometimes better than Linux.
John Wiltshire
Fear: When you see B8 00 4C CD 21 and know what it means
There's a review of web server platforms in the same issue that disagrees about how wonderful Linux is.
Seems like they 'forgot' a couple of ports (maybe to make NT look less bad?) like the Corel netwinder (ARM), palmpilot (like someone else said), SGI's MIPS systems. .. see http://www.linuxce.org/ linux as a windows CE replacement on MIPS/SH3/(ARM maybe?) devices. -/plug-
.. wasn't it an improved version of 2.0.x ?
And -plug- support for more is on the way
Their "table" said that NT4 supports IP masquerade, yah right.. win2K and the latest win98-osr2 ? support it (but who knows how good it's going to be).
They also claim that linux 2.2.x adds Raid support
Dork.
Regarding installation difficulty, at best they are both roughly the same -- but NT does a much better job of working with unusual or new hardware. Linux, particularly RH Linux, beats the tar out of NT when it comes to disk management though. Unfortunately both suck rocks at PnP support, can't give either one the advantage here.
... not much of a contest, particularly given the price differential.
Regarding "user-friendliness", I find the opinion that X is more friendly to be hard to support (and I've been using X since before Windows 3.x even hit the streets). Let me explain.
The variety of window managers and desktops makes X a lot harder to deal with for the beginner; most of the time you don't just get a working desktop when you log in and start it up, you have to tweak the hell out of it. Red Hat and Caldera have made great strides here, but the fact that there's no standard sure makes things tough.
That assumes that the server even starts up. One huge, huge problem with the design of today's X servers (at least the ones I've used) is that they tend to be monolithic (server tuned for a particular video card or class of cards). They should be redesigned with a loadable driver system instead, and a hell of a lot more work needs to go into getting them to properly identify display systems.
Don't get me wrong, I'm not particularly fond of NT, but in terms of end-user usability it creams any UNIX variant I've ever used (even NeXT, although that had a few high points).
Don't even get me going on support for peripherals such as printers. They scale well, but they can be a freakin' nightmare to configure even for an experienced UNIX hand.
If you're not focused on UI issues, though, Linux generally whips NT's ass. Gateway services, email, web services, file sharing
jim frost
jimf@frostbytes.com