Slashdot Mirror
Three Enterprise Operating Systems Compared
Anonymous Coward writes "Finally, a much awaited review of enterprise OSes. The guys from NW Test Alliance pitted
Red Hat, UnitedLinux, and Windows against each other and rated them on several rubrics. Red Hat won by a slight margin on the basis of its high hardware compatibility and strong security integration."
This isn't a Red Hat vs. UnitedLinux vs. Windows review. The declare Red Hat the victor over UnitedLinux. The compare some things, such as max tcp connections and file transfer times against Windows, but never do they declare that Red Hat has better hardware support or is easier to configure than Windows.
get nemulator
Even though it would be "fun" to see a comparison between Linux and Windows, I don't think it really could and should be done. Mr. Gates and Company would like for us to think that it is a viable solution to everything but honestly, as we all have discovered, there is no silver bullet. So what Windows may be good at something Linux may suffer at and vice verca. Now to know each ones strengths is truely valuable.
However what the article does with the two linux distros is good. Now we are comparing two OSes designed for the same general tasks and let them duke it out.
But in the end, I would like to see some list of strengths.
That's true enough, but if you're designing an enterprise system, you're going to want to use whatever's best.
A more comprehensive set of tests may have shown that, in fact, Windows 2003 Server is best, at least ignoring cost, licensing, etc. Without making this "apples and oranges" comparison, you don't know.
I support open source as much as the next person, but I also support using the best tool for the job.
It's official. Most of you are morons.
Enterprise distos are all about clustering and load distribution, but these tests are caried out on single machines. What is the point?
I'd like to see comparisons on a number of purposes (webserver, database server, etc) done on a range of identical hardware (low-end, mid-range, and high-end servers, or as equivalent as possible in the case of OSX) where the OS installations are done by technicians who are intimately familiar with their particular OS. I'd like to also see these stats updated from time to time with significant new releases from each company.
It bothers me when I see people with a whole lot of experience on one OS and some experience on another OS criticizing something about the one in which they have little experience, and this applies in any direction. As one who has far more experience in Windows than in Linux, I wouldn't expect to be able to set up a well-configured RH web server (working on learning), though I could probably get something basic in place. I've seen the reverse when dealing with Unix people, who have difficulty understanding some of the ways in which Windows handles things.
So far, most of the tests I have seen have either not been comprehensive enough, or have been slanted by the bias of the testing group. I've seen few examples of tests including OSX server, and it would be nice to see how well some OSes scale *down*, since not everyone can afford a $10K or more server for their first foray into whatever it is they want to do.
You can never go home again... but I guess you can shop there.
No, Debian is not enterprise ready. To be enterprise ready they need ISV and OEM support like RedHat has but more importantly, they need a company that would provide enterprise-class support AND release engineering for the OS similar to what RedHat does with their AS/ES/WS product line.
Why didn't they include any Enterprise Operating Systems in their comparison of "Enterprise Operating Systems"?
I mean, like Solaris or AIX.
What is this reviewer smoking? Statements such as,
The wizard worked well and mostly made astute choices, although it divided our disk arrays into seemingly bite-sized devices with seven partitions. By contrast, the UnitedLinux distributions divided the two disks we used into larger chunks, which is a better way to reserve server space for future operations.
Shows that Tom Henderson doesn't know what he's talking about. How could anybody think that one large partition is better than lots of smaller ones. If one is consentrating on enterprise level systems one would be using LVM and have lots of partitions so they could add drives as they go and increase the partition sizes on the fly.
This isn't to say that the conclusion is wrong - it may be entirely correct. It's just to say that I get pissed off by pointless "scoring systems" that are apparently objective (they're numbers...) but are actually completely subjective and just intended to give a spurious authenticity to the conclusion. If they said "We think Red Hat's security is better and that's a reason to prefer it", fine.
And if you don't understand why a result based on a scoring system where the difference in scores is less than the expected uncertainty of the result is not valid, then what are you doing trying to benchmark a technical product?
Oh well, rant over for now.
Panurge has posted for the last time. Thanks for the positive moderations.
Quite so. Webserver, I can just tell you, Linux will walk away with,
but I'd be very interested to see such a comparison for the database
server, (SMB) fileserver, thinclient server, and whatever other
categories the people organising the comparison think important.
(Print server is probably not necessary any longer, at least not
with the high-end hardware, now that you can get a really nice
network printer with a full maintenance contract and also use it
as a color photocopier on the side... but I'm sure there are
other uses for which the comparison could be done.)
> It bothers me when I see people with a whole lot of experience on
> one OS and some experience on another OS criticizing something
> about the one in which they have little experience, and this
> applies in any direction.
I have more experience with Win9x than any other OS, but I criticise
it more than any other OS except pre-X MacOS. Actually, in general,
I tend to criticise OSes in direct proportion to how much experience
I have with them, because it's by experience that you learn the
foibles, the things that are _wrong_ (not just different) with an OS.
I switched to running Linux full-time on my desktop about a year
ago this past April or so (though I'd multibooted for a while before
that), and I'm getting now a pretty good feel for what's wrong with
Linux (or, at least, with Mandrake).
> So far, most of the tests I have seen have either not been
> comprehensive enough, or have been slanted by the bias of
> the testing group.
Indeed, and that goes both ways. Microsoft pays some "Research"
group to prove NT is better, and then the Linux blogs post stories
showing that Linux is better, written by Linux geeks. I don't
trust either side of that. And then of course Apple will tell
you that Mac OS X is the best; it might be a _little_ easier to
believe they know what they're saying if they hadn't said that
about Mac OS 8 and 9 too, which didn't even have multitasking,
but even then I'd still rather hear it from someone who gave
each system a fair shake.
And yeah, I'd want proponents of each OS to configure that OS,
and then the people doing the judging to compare. Either that,
or all three OSes should be left in their out-of-the-box state,
in which case it might matter deeply which distro is selected
to represent Linux.
Cut that out, or I will ship you to Norilsk in a box.
I thought the title said Enterprise OS. All of the >$10 Billion/year companies I've written software for run *nix on Sun and/or *nix and/or an IBM OS on medium to big iron. They are not running Windows as an "Enterprise" platform.
I'm not talking email servers where a few poor sap CIOs got talked into running Exchange farms, or similar unfortunate tragedies with IIS, I'm talking the ERP stuff that runs the factory, accounting, payroll, and other stuff people have to bet their businesses on.
I realize OS/390 and Windoze are apples and oranges, but come one, they said ENTERPRISE. Now if they mean "Enterprise" as 2 guys and a van and a laptop, then hell yeah bring on the Windows. Otherwise, it's like having a review of the world's fastest street cars pitting Acura vs. Mazda vs. Toyota. The Lamborghini and Ferrari folks are tapping their feet and rolling their eyes. Put DB2 on an S/390 and on the bitchinest Windows box you can get your hands on, then do the test. I dare you.
"Stop throwing the Constitution in my face, it's just a goddamned piece of paper!" - George W. Bush Nov. 2005
WTF? Nice to see that the Novell was once again left out of the testing. Why don't you Linux Zealots try and broaden your horizons. After all the recent Novell is "Linux's best friend" posts the last couple weeks and still they get no respect. Novell would rape your Linux in such testing. Also Novell is now giving away 5 user Small Business Licenses. You have to jump through some hoops to be able to get your hands on it, but it is pretty painless. Novell is by far the best NOS out there, it is mature, stable (600+ day uptimes any one), and has great applications. Also most if not all on Novell'a apps run on UNIX, Netware, Linux and Windows.
For the love of god Linux is not the end all be all of NOS, if you hate M$ that much (I do) look at all the alternatives. Free does not make it better.
Friendly