“As good as Mac OS X is for desktops and laptops, one wonders if the FreeBSD inside is not too restricted by the Apple jacket around it to also make for an efficient, secure and fast server OS. Apple is now busy convincing the world that Apples make also for excellent server appliances in the handy U1 format, thanks to OS X. That new product is called Apple Xserve. Many potential buyers are, however, asking themselves if OS X—given its recent introduction—is ready today to handle their critical apps.” Read the article at Byte. In the meantime, Apple released Darwin 6.0.2 for both PPC and x86.
Comparing Apples and Penguins
2002-10-29 macOS 51 Comments
Anyone knows if Darwin can have its /boot on the same partition as the /root, as most modern Linux versions can, or does it really require a /boot? I only have one partition available to experiment with it.
Could this reviewer use a more out of date set of tools? 10.1.5? Kernel 2.4.18? C’mon, get with the program! Mac OS X is up to 10.2.1, based on FreeBSD 4.4 while the default kernel on most Linux distros is 2.4.25. Sheesh!
>Could this reviewer use a more out of date set of tools? 10.1.5?
I think he used whatever the XServe came with. That is the right thing to do. You review with the default installation.
>the default kernel on most Linux distros is 2.4.25
And now, that shows how you don’t have a clue of what you are talking about. The latest stable version for Linux is 2.4.19 with the latest 2.4.x unstable one being the 2.4.20-pre11 (http://www.kernel.org/). The 2.5.x versions of Linux are unstable betas and this source tree is going to be released next June.
Even Red Hat 8.0 came with 2.4.18, it is a very recent kernel.
As was debated last week, the OS X ABI appears to be less than optimal as well, and could very well be the source of the some of these slowdowns.
The real question, is what does the X-Serve buy you running OS X vs the X-Serve running Linux. Does it have better integration with the hardware itself, and for administration than the Linux software provides?
That is, if you were running a solely Mac shop, is the X-Server easier enough to administer and operate using the stock software than a PPC Linux would be?
I don’t know, as I don’t really know the feature set of the X-Server sofware.
You’d like to think Yes, but it’s the most interesting question I could come up with.
Anybody smell Apple retreating to the x86 platform?
It is sort of a shame he couldn’t have used OS X Server 10.2.1, to see if the results would be different but, if 10.1.5 is what came with it, then that’s what you use.
>Could this reviewer use a more out of date set of tools? 10.1.5?
I think he used whatever the XServe came with. That is the right thing to do. You review with the default installation.
Unless you want to intentionally skew the review negatively. For that you change the default desktop to something totally random, add an app that isn’t in the default install, and then complain about how bad the OS looks and behaves. (cough…Mandrake 9 review…cough….)
Changing the background of a desktop (after a whole week of using Mandrake in its default install), hardly changes any performances. The Mandrake 9 was reviewed just fine. And MandrakeSoft could not rebute any of the things I wrote in the review. So, cut the sarcastic crap over here, or you will do get banned.
HE said in the articel that he got the server form a friend, that server probably did come with 10.1.5. But what I don’t understand is doing a review with an older OS. The reason Apple updates the OS is to get better performance. I should do a review of 10.2.1 and something like Mandrake 4 (?). I’m sure OS X will win in that case.
Depends how you see things. MacOSX 10.1.5 was released only a few months ago. It is a new OS version, no matter if Apple in the meantime had released a newer version.
On many Windows OS/dev comparisons, Win2k is used instead of WinXP. But people still think the results as valid.
Obviously, the reviewer did not want to provide an up-to-date review. No XServe has been shipped for a while without an OS upgrade path included in the box. If he chose to go with the default configuration, fine – then he should note that this isn’t the *currently shipping* configuration.
It’s Apple’s fault in that the upgrade cost so much that it’s inconvenient to upgrade. The reviewer may have been inclined to do so if it didn’t cost so much. Sad, but true.
However, his review is a bit… odd.
We can’t say he wants to do everything with a stock install, ‘cuz he doesn’t. He makes the mistake of compiling all software with gcc 2x, and even cleanly compiles gcc! With all reasonable optimizations. If he’s reviewing for folks that do that, he might have wanted to apply some custom patches from the darwin source tree to improve server performance, and then used the latest gcc, then not try to guess what the correct optimizations are…
I read it, and it says what I expected – Mac OS X 10.1.x sucks. Or that should be the past tense. It sucked. Jaguar sucks less – a current review (in an October dated issue, at least) should do as Eugenia so often does – expose everything that’s being done. And review consistently – either as a power-user, unix-geek, server god or some amalgamation of the aforementioned choices. Gah, aforementioned is such a pompous word – sorry Seriously, though – as a true server-type dude, I don’t touch fink w/a 10-foot pole. It’s nice, but I prefer to get the source and beat it with the pole ’til it works. That way, I’m in control.
All this has been said to say his review is technically correct. It’s what we’ve all heard before. And I’m waiting ’til 10.3 comes out so we can see how Jaguar does against the latest linux kernel/good distro. ‘Cuz by then, the *real* server software (Maya, lightwave, mathematica, sybase, oracle, etc.) will be revved and up an’ blasting away. Then it’ll be good.
This from a person who hopes Apple makes it ’til the chip rejuvenation. Beautiful tech, magnificent middle management… odd strategic moves.
Yes, I think odd is the word.
As always, thanks for reading.
“And now, that shows how you don’t have a clue of what you are talking about”
I have some clue:) I mistyped.
I expected OS X to outperform Linux, or at least do a lot better than it did… Apple tends to be really paranoid about giving out hardware info, so I figured OS X would just be faster because it would be better optimized for the hardware.
So it makes you wondering that Linux managed to outperform OS X on it’s native platform (not a troll, I’m being sincere).
I agree with Eugenia 100%, the default configuration should be the one reviewed, though it might be interesting, that in addition to these tests, that a more recent version of Mac OS X and a tweaked Linux were added to the tests (ingo’s scheduler and rmap).
FreeBSD 5.0 has been ported to PPC right? It will be quite interesting to compare OS X to FreeBSD itself… (yeh yeh, 5.0 is quite an improvement over 4.* but a comparison would be interesting nonetheless).
The results of this benchmark really were a foregone conclusion. Linux is just plain fast. Even years ago when it wasn’t super-stable, it was still fast. Benchmarks like this spend a lot of time executing standard UNIX system calls, the speed of which are measured quite well with something called lmbench.
In the lmbench tests, Linux (and NetBSD) absolutely smokes Darwin. You can find the results at: http://clustermonkey.org/~laz/pbook/rob.lmbench.txt.gz
The tests are based on Darwin 5.1 (OS X 10.1) but if you’d actually take the time to read the changelogs, you’d realize that very few changes were made to the core OS. The stuff that made 10.2 faster were in the GUI, which wouldn’t really affect these XServer benchmarks.
The same old people here are always whining about how horrible OS X performs, and here, it holds up well to Linux in a very unupdated form. Then, there are those who’ll say: “…but if you’d actually take the time to read the changelogs, you’d realize that very few changes were made to the core OS. The stuff that made 10.2 faster were in the GUI, which wouldn’t really affect these XServer benchmarks.” as if they are informed… Oh? Moving the core of the system from v3.1 to 4.3 won’t show improvement? Recompiling with gcc3.1 won’t help? Since this is a test of a pure server benchmark, running Darwin instead of a OS X wouldn’t provide an improvement?
Baloney! Let’s see these tests run again in a month.
The only thing that surprised me was how negligible the difference is under these circumstances, and that Moshe seems to know hardly anything about OS X/the Mac.
He’s better off than half the fools here, and certainly better off than the Slashdot crowd, but, boy? Are newbies required to live in awebless, lifeless vacuum to get acquainted with X or something?
>>>> I expected OS X to outperform Linux, or at least do a lot better than it did…
as a mac fan on the desktop for some folks, and it being my first computer, i would be more then happy to see that outcome.
if you are talking strictly server performance with linux running on x86…(and os x running on xserv of course) it’s not even a contest. and never will be.
linux sitting at a command line, with about 5 minutes of optimization running on decent x86 hardware will stomp the crap out of mac anything, when it comes to number crunching. ….for years to come.
people don’t buy macs to be the fastest anymore…you need to get that out of your head…macs have some good selling points. Being the fastest ain’t one of them.
if you wanted to test yellowdog versus os x on the very same machine….theoretically if terrasoft did their homework, linux should win everytime.
there’s like ZERO overhead running linux. just load the bare minimum and the daemons of choice.
how can osx even compete with that?
again, i’m not trying to say mac sucks…i like macs very much. i’m not trying to say that macs don’t have redeeming qualities. they do. very much so.
get it out of your head that macs are fast.
they are far from the fastest.
i’d call them “reasonably” speedy.
now cut that “expecting” out, before you hurt yourself.
by the time the baby Power4 comes out, amd and intel will be mopping it up.
OS X is based on an ancient codebase (Mach 3.0 and 4.4 BSD) that hadn’t been worked on for years until Apple came along. Even afterwards, Apple had a (comparatively) small group of people working on the code. It’s got a schizo kernel design that makes a monolithic kernel run on top of a (crappy) microkernel.
In comparison, Linux is a clean modular kernel design that has had dozens of hackers working on it for the better part of a decade. It has the latest algorithms and uses the latest optimization techniques (per-CPU global data, cache-aligned variables). Why, oh why, did anyone expect OS X to be faster?
OS X has it’s good points, even in the UNIX layer. It’s XML configuration stuff, in particular, is great, as is Core Audio. But speed is not one of those good points.
PS> In Darwin 6.x, only the FreeBSD portions were updated to to 4.4. The FreeBSD portion is mostly the user layer and parts of the filesystem layer. The updates would certainly help, but wouldn’t come close to briding the performance differentials shown in the lmbench numbers.
PS> Also, updating to GCC 3.1 probably wouldn’t help much on these. Like I said, these benchmarks are very dependent on the performance of the kernel and its algorithms. For this type of work, code optimization matters very little. Think of optimizing compilers are making small runs of code faster. If a task consists of a small bit of code running in an inner loop (most imaging and multimedia-type code) then a better optimizer helps a lot. If, instead, a task touches long regions of code and manipulates many different types of data (like server tasks do) then the micro-optimizations performed by the compiler take a back-seat to the algorithms themselves. Besides, if GCC 3.x helps the Darwin kernel, then it’d help the Linux kernel equally well, and the performance difference would remain.
PS> As for better optimized for the hardware, the only undocumented part of a Mac is the stuff on the motherboard. Manipulating stuff on the motherboard tends not to be performance intensive. The important stuff (IDE interface, PCI bus, AGP bus, etc) are openly documented and equally easy for anyone to optimize.
I’m not suprised that Linux outperformed Darwin even on an XServe. The Linux kernel is very fast and the Darwin kernel is pretty weak and rumor has it is still mainly optomized for 68030.
However you don’t buy an XServe for value per $. What you are buying is ease of administration and the tool kit. There is no question that Apple doesn’t offer the best horse power for your money; what they often offer is the best level of employee productivity for your money. I think a test where the XServe could really show its merit is by asking so / so administrators with no Linux or Xserve experience to perform say 25 varied tasks (things like reconfigure apache for two virtual websites, set the mail server to copy all outgoing mail to external addresses from specific accounts, reconfigure the network setting on 100 client machines..).
I certainly would agree 10.1.5 vs Linux would be a foregone conclusion, but I wouldn’t make predictions about 10.2 vs. Linux. While the GUI certainly was tremendously improved, the rest of the OS runs quite a bit better.
Here’s some snippage from a usenet post I made a while back, comparing the performance of postgresql on the same hardware:
I ran several iterations of the postgresql 7.2.1 “runwisc.sh”
benchmark on an 800Mhz Apple TiBook with 512Mb RAM. The first run was
compiled using MacOS X.I, the second set was the same executable on
MacOS X.II, the third was a recompile using the GCC 3.1 based compiler
shipped by apple (Build 1151), and the fourth was another GCC 3.1
recompile, but this time with -faltivec .
Version Mean clock time
MacOS X.I w/ gcc2 : 22.06 seconds
MacOS X.II w/ gcc2 : 13.52 seconds
MacOS X.II w/ gcc3 : 13.16 seconds
MacOS X.II w/ gcc3 -faltivec : 13.31 seconds
My only hesitation in trumpeting the virtues of X.II vs. OS X.I is that I only ran the benchmark once on MacOS X.I, as opposed to multiple tests of the other configurations. I also ran it in a “Terminal” window in Aqua, so it is possible that it was just the old crappy GUI getting in the way.
Eugenia, either I don’t get you right or you’ve seriously gone astray here… do you mean by reviewing the default that I had to review NT (haha..) or W2K without service packs if I felt like doing so right now..? I hope that is not what you tried to say. Of course one has to always review the latest stable version, there is no point arguing about it.
So, you all knowledgable guys who have an opinion on everything, I asked a very simple question in the begining of this thread.
No one answered yet.
I do not believe it would, but your welcome to try it. I’d take it that their silence means non of them run darwin. I know I don’t.
I can’t help you with your question, but why don’t ask you the Darwin-mailing-list ?
Well, I am certainly no mac fan. >:)
But I guess this is like the case of how Linux outperforms SunOS 4 and Solaris on sparc (yes it’s true, I’ve seen it in person). Well, I am sure this is not the case with many-processors+solaris of course.
I don’t think Macs are fast, I wonder why you brought that up? I don’t think it’s relevant since Linux and OS X were ran on the exact same hardware. (or were you also just making these statements to anyone?)
As for compiling with gcc3.1, gcc3.1 is available for linux and has been for some time, you could just compile the linux kernel with gcc 3.1 (yes, there’s 3.2, but it test results might be a bit more fair if the same compiler is used is used with each OS). So compiler benefits of gcc 3.1 for darwin are there for linux as well..
As for better optimized for the hardware, the only undocumented part of a Mac is the stuff on the motherboard. Manipulating stuff on the motherboard tends not to be performance intensive. The important stuff (IDE interface, PCI bus, AGP bus, etc) are openly documented and equally easy for anyone to optimize.
Well, I was thinking more in terms of optimization for the G4 but ok.
I have a Sawtooth G4 that has been overclocked to 533, I have been using Mac OS X since the first release.
The first release was horrible for speed, everything lagged and the system took forever to launch a simple text editor and browsing the net just wasn’t acceptable with the amount of lag I saw.
With X.1 the speed enhancements were evident and I would agree that 30% faster was the minimum performance boost I saw.
But X.2 Jaguar is a whole different ball field, the system is far more reactive, programs launch with ease, and now that I have a Radeon 8500, I never see video lag, plus my CPU load on idle is less than 6%, without the features of QE it’s about 20%.
OS X and this “slow down” is no longer evident on my system, and it’s on of the first G4’s ever released, if you are still complaining about the speed of OS X I recommend upgrading to Jaguar. Besides everybody knows Jaguar was compile with GCC 3.1, which provides many performance boosts, this is why I believe the first post was saying 10.1.5 was an unfair test. People should take these comparisons lightly, for they are not accurate anymore although I believe it is important to test the default system that came with the Xserves because most people who own them probably aren’t too keen on upgrading them.
I tried to download Darwin for Intel and the site keeps on asking me for a username and password. Can Anyone help? Please. I want to try it out soo bad since I don;t own a Mac.
Yes for now one needs 2 partitions in order to boot darwin x86.
Some patches are availaible that make darwin x86 work correctly on AMD based machines (patch available somewhere in the gnu-darwin project).
For those of you wondering why Darwin x86 lags so much behind, the reason is simple, Apple does not need to upgrade hardware support, Apple just needs to verify that it’s code is cross-platform. Making a broader support for x86 hardware should be brought by the non-Apple darwin community.
Apple hosts http://www.lists.apple.com/mailman/listinfo/darwin-x86 a mailing list dedicated to x86 support in darwin. As you can see on the list archives http://archives:firstname.lastname@example.org/mhonarc/darwin-x86/, trafiic is low meaning poor support and small number of updates for the x86 specific parts of darwin.
If becoming an Apple registered Open source developers bothers you and you don’t want to join the Apple Hosted x86 initiative you can check out the sourceforge hosted gnu-darwin project which also works on the x86 part of darwin ( on the contrary Fink does not). But If you want to make sure your contribution will get back into Apple’s main darwin tree source I do suggest you become a member of http://www.Opendarwin.org . There is no x86 project has of now but one can always ask for such project to be created. In the meanwhile patche can be added to bug reports at http://www.opendarwin.org/bugzilla/ .
Because you need to become an Apple OpenSource developer which is not the same has being an ADC member (see http://www.opensource.apple.com/APSLRegInfo.html ).
With this account you’ll also be able to browser via the sources via cvsweb and also be able to co the sources. If becoming such a developer bother’s you, you’ll need to wait for gnu-darin and http://www.ptf.com/ to make CDs available for purchase.
Eugenia, why don’t you try darwin on your cube ?
My Comp Sci profs were very critical of MacOS X’s core, but I’m surprised at how close it came. There are only 2 other projects that use the Mach microkernel that i’m aware of, GNU Hurd and Windows NT/2K/XP. That doesn’t instill confidence in me.
I can’t wait to see the new I/O subsystem of Linux 2.5 when it comes out next summer(fingers crossed). Supposedly its faster than 2.4 by a fair amount.
>Eugenia, why don’t you try darwin on your cube ?
Because my Cube has no available partitions. It came already parrtitioned as OSX and OS 9, on a small 20 GB hdd, so no way I am going to f*ck up that installation just to install Darwin. I don’t have the OS 9 CDs anymore anyway, so any changes I would do, would mean bye-bye OS 9. I won’t do that.
I am only open to try it on my x86 machine which had a partition available for that stuff, nowhere else. But I am not willing to repartition that disk which has 7 more OSes installed, too much of a hassle and a danger (if Darwin gives partitions numbers the way Fdisk on linux does, I risk to lose BeOS).
Better safe than sorry. I will wait for Darwin to allow such things, and then try it.
“so any changes I would do, would mean bye-bye OS 9”
What do you use Macos9 for may I ask ?
Nothing particular. I just don’t want to repartition and have to reinstall all that stuff, just for Darwin’s sake.
Don’t try to talk to me over it Ludovic, I don’t want to touch my Cube. I like it the way it is.
If I will install Darwin, it will have to get installed on my Celeron box.
In fact, as I am writing this, I installed Syllable on the partition that Darwin would get installed. I will wait until Darwin is not as demanding.
You can exit Aqua by logging in with username “>console”.
And I really wonder why he installs “OS X” on Server hardware (instead of “OS X Server” of course!)
Performance is not all there is to servers. Contrary to Linux, OS X Server has a set of very mature configuration and monitoring tools which really speed up system administration.
When compared with X86 boxes and available OSes like BSD, Linux, Solaris etc… The price/performance is a no brainer.
Linux Blows OS X away and it’s free…
X-86 Servers are overall faster and less expansive…
More OS options and more apps…
AU/X didn’t make it and neither will XServe. There are just too many good free Unix server OSes out there. At least Apple has a server to go with their desktop so I guess it’s OK… …but sorry, I ain’t buying!
Eugenia, sorry if this is a dumb comment, but if you’ve got MOX on your Cube, you already have Darwin installed, don’t you? Go into console mode (e.g. from the login screen) and you’re on Darwin. But maybe I’m missing the point here entirely …
I have to save though, even though if he uses Jaguar, Linux would still beat it hands down in performance (okay, I admit, I have only tried 10.2.0… big deal). Now, just say you dump the expensive XServe and get a HP or Dell…. or let’s get radical, a IBM. I would get better performance for the buck 🙂
So….. from what I see, the only way Mac OS X can beat Linux is ease of use? Sad, sad.
How much is the upgrade price (in terms of cash) for downloading SP3 and installing it? How much is it to buy a new Jaguar pack? Comparing free updates and real hard cash updates is bad. If you want to do that kind of comparison, use OS X 10.1.0 and Windows 2000 5.0 🙂
not exactly, you have more binaries. Pure darwin is set up a little differntly which is a pain in the **** when one needs to compile xnu on X.
But it will take a while to reach the speed of xBSD/Linux.
All in all though, the benchmark results were not abysmal or anything, be realistic folks.
The kernel in OSX has unfortunately nothing to do with FreeBSD and it’s a pretty old design.
The userland stuff was upgraded in Jagwyre but that’s not the whole story, it’s funny how many people still think OSX is a pretty recent version of BSD.
In a year from now OSX will be better, hopefully they’ll be running on decent hardware by then and they will have implemented a decent kernel and filesystem.
These same Comp Sci profs were criticizing Linux a decade ago because it wasn’t Mach That said, GNU Hurd uses a much more recent version of Mach, and Win 2K never used Mach at all. NT is kind of a microkernel. Originally, it was a fairly clean one, with a Win32 server sitting in userspace. Now, the Win32 server is still seperate (crss.exe on 2K and XP) and communicates via IPC, but is in kernel space. The microkernel itself is not Mach at all, but a in-house design influenced by VMS.
As for 2.5, I’m running it right now, and (for desktop stuff) it does seem a little faster. The main thing is it behaves nicer under load than previous kernels, and has the benifets of preempt built in. Apparently, Linus himself is quite happy with the new VM and I/O systems.
What you pay for in the mac is ease of use. If you have less time then money then the XServe is worth it.
I cannot begin to tell you how I have been tempted to push my linux boxes out of a window because I cannot install something because something is missing and i can’t find it.
I won’t even begin to talk about the Xwindow problems I have had.
Sorry but I want to turn my computer on and have it work, not needle around for hrs to try and make it work.
It’s a fine review strategy to use the tools that were actually in the package with the item you are reviewing.
But XServe ships with OSX Server 10.2 now (and for some weeks, if I remember). Given that some of the key improvements in 10.2 are to speed and networking, it seems this review is irrelevant. It says virtually nothing about an XServe that one would purchase today.
As good as Mac OS X is for desktops and laptops, one wonders if the FreeBSD inside is not too restricted by the Apple jacket around it to also make for an efficient, secure and fast server OS.
The FreeBSD what… userspace? The FreeBSD components in XNU merely replace the functionality of Mach servers… the important kernel components from a performance perspective (VMM, process management) all come from Mach.
Apple sold 5,700 Xserves last quarter. From “zero” position in the US server market six months ago, to now they are the number 5th computer server maker. Not bad. I guess you should ask their customers why someone is buying Xserve.
My guess is they are Education and Biotech customers.
The people here who are wondering why they would want to buy an Xserve are the same people who are wondering why anyone would want to use a Mac for anything. So I don’t understand why their comments are even relevent to the discussion. The reason to buy Xserve is gain the capability to run Mac OS X on a good server platform from Apple. If you want to run Linux on an x86 box, then do so. No one’s forcing you to buy an Xserve. Why must people continually feel that any product that comes out on the market must be somehow be either defended or attacked? Can’t we just all be friends?
<quote>If you want to run Linux on an x86 box, then do so. No one’s forcing you to buy an Xserve. Why must people continually feel that any product that comes out on the market must be somehow be either defended or attacked?
Listen I am there with you man. However, most of the comments seemed to be critical of linux and pro-apple as in “If they had just used 10.2 it would have kicked Linux butt.”
There is a serious place in Mac organizations, schools and other institutions for this kind of server. Now, there is usually no need for small organizations to have servers running different OSes on different hardware. The Xserve has its place in the server food chain to be sure.
There are people who are always going to buy big unix iron for their server farms. There are others who choose BSD over linux for what they see as better security and such out of the box. Then there are people that understand what to tweak and how to get blinding fast linux box.
The key is to understand what you want and how to get there.
The funny thing is that OSnews covers more linux than just about anything else but I swear from the comments that most people that visit the site dislike the OS in general and treat every person that does prefer the OS as if they are all slashdot trolls on the prowl.
For example, I will make a bold statement.
Linux was initially designed as a desktop Unix-like OS for a geek that wanted such an OS but hated some things about Minix. It has been used as a desktop OS from the start. However, it was created by geeks for geeks and that is the way I like it. All the choice you get with BSD or Linux on the desktop comes at the heavy price of having a system you have to configure heavily before it “just works”. If you don’t want to get on the command line and get your hands dirty a bit then please don’t bother.
I am not a knee-jerk MS hater. Use XP it came with you x86 box and its pretty darn stable and it has lots of apps. Just use it. Stop trying to install an operating system that by your own ommision you will never like because it will never be just like Windows and I hope it never will.
If you hate the MS way of doing things then start saving up money for Mac.
It took me a couple of hours of casually tooling around but I have my Dell 4000 Inspiron with Redhat 8.0 installed, lucent modems working, Fn keys for adjusting sound working, Xft (smooth fonts) version of mozilla with all the plugins I want, MS core fonts, mp3 all multi-media, windows compatibility with wine and the gtk2 versions of the software I want. It would have taken a newbie days maybe weeks to get this far.
However, I have seen Windows newbies spend days hosing around with their machine finding drivers and such after XP came out. I have seen Linux newbies get their box running with just the defaults and never even think about half the stuff I listed above.
It all depends on what you want.
I totally agree with you. I use Mac OS X as my primary OS now, but I’ve toyed with Linux for desktop use in the past, and I’m very impressed with the strides made in usability and the like in the open-source arena. There’s plenty of room for all OSes in both the desktop AND server worlds, and these stupid little platform wars continue to accomplish nothing other than sucking up bandwidth.
“The people here who are wondering why they would want to buy an Xserve are the same people who are wondering why anyone would want to use a Mac for anything”
simply not true.
several of my friends make up a small group that are mac, pc and linux proficient.
we all own and run macs, PCs loaded with windows and linux (and freebsd in my case).
there are many instances where i will suggest a mac.
i won’t be suggesting an xserv anytime soon.
my friends will agree with me.
you see, it’s because they are not trapped on one platform, and have blind advocacy for one platform.
and that’s the reason why we won’t suggest an xserv.
they work for research facilities with a high percentage of macs. they specifically work with databases and php….they run os x on one desktop, and redhat on the other.
linux is on the servers.
i dabble in graphics work…i use macs for adobe and macromedia apps.
i also use redhat as my 3d workstation of choice even though the same app was available on a mac.
i setup dns/web/mail servers running on redhat.
so you guys need to quit categorizing everyone.
1.mac zealots say xserv is good
2.linux zealots say xserv is not good
3.windows zealots say xserv is pointless
4.mac/windows/linux proficient person states xserv is interesting but would not recommend one…except for a small “all mac” shop.
you tell me which statement might contain a bit more truth then the others. and now you understand why i consider most of the posts about the xserv not very objective.
The article plainly said that this was not a supposed to be a super accurate test. He has a disclaimer right from the start. So witht hat in mind I think OSX held up well. I dont understand why he did not use OSX Server.
Ya I like the concept of LINUX but the ease of use and support of OSX keeps me in that arena. Now only if Games were made for OS X , I guess I am stuck with my cheap fast Athlon for gaming and everythign else my iBook.
I would like to see a test of the “best” LINUX and the Latest OS X Sever. I am sure linux would win but I would be more interested in how close they really are. And I wish they would throw in a MS box for fun 🙂
I have a few question and comments. First, I have not work with Linux recently; does one still set up a separate swap partition in Linux?
I know in OSX that the default install does not use a separate swap/vm partition. Swap/VM is uses free hard disk space.
In the OSX discussion groups several people describe how to setup a separate swap/vm partition one a single drive for a significant performance improvement.
I suspect Apple has decided to keep things simple for the “average” user and let the OS deal with swap/vm on a single partition.
It would be interesting to see the test repeated with a separate swap/vm partition