Slashdot Mirror
Microsoft Challenges Linux's Legacy Claims
Michael writes "Microsoft Corp.'s Linux and open-source lab on the Redmond campus has been running some interesting tests of late, one of which was looking at how well the latest Windows client software runs on legacy hardware in comparison to its Linux competitors. The tests, which found that Windows performed as well as Linux on legacy hardware when installed and run out-of-the-box, were done in part to give Microsoft the data it needed to effectively 'put to rest the myth that Linux can run on anything.'"
When I read the blurb, I figured out what I would find in TFA. They're comparing XP to, say, SUSE 9.0 or RH 4.0 EL. Both optimized for current systems. Here's the difference; there are many distributions of linux targeted at older, slower machines, going all the way back to 286's. I would like to see performance comparisons between Windows and a linux distro targeted at smaller machines. See distrowatch; they list a couple that are *meant* for this application, so you don't have to be a kernel hacker or techincal expert to modify your linux distro for your hardware.
*yawn*. Same old MS crap.
Thinking outside my Head
Right, I am wondering about the subjective nature of this article. I have found that linux runs great with the scarcest resources. Tell me where you can run a full PBX and IVR using a P-II 300 mhz? My Gentoo + Asterisk did just fine. Just my $.02, I am afraid that this is going to turn into another *nix vs. windows argument. I think that every OS has a place on the network. Just the thought that Windows 2003 was installed on a PII-300 makes my shudder. However, Linux is quite happy. (As well as your favorite BSD's)
I don't know what they were installing, but not the distros I use. See... lesser known fact about *nix is that it comes in many flavors. If, say, you had an older, piece of junk, you can get just as new a version of Damn Small Linux as you could Fedora Core 4. One is 50 MB, on a cd, the other 6 GB on 4. The thing they're assuming here is that you have to have a GUI to be productive. I call shenanagins.
I've done this same test with a box I 'liberated' from another source. (Was given to me, as it was too old to donate, believe it or not). 'Tis a first gen Pentium, with a whopping 32 MB RAM. I've got Fedora Core 4 on there just fine! It works as a web server, a file server, as well as a programming workstation, and email. I even browse the web on it fine! Oh, one small thing, it won't run X, de to size. (Ok, it will, ya just don't wanna... trust me). Guess what OS was on there previously? a very, VERY sluggish version of 2000. I don't know how they kept it running, but they did.
I ask you, which is better on legacy hardware? The ability to choose what you need, so as to maximize what you have? Or the ability to run everything in the world, and see what breaks?
To the people out there about to mod me flamebait: Yes, I read TFA, and no, I don't buy it. To judge to world of Linux on a few distros is foolish. Just as they test a bunch of versions of Windows, they need to do a range of Linux. Jump to the end of the article:
Good, they acknolwdge what I just said. But again, how do you define out of the box? Is it whatever boots from the CD? Or a 'full install'? I really think this is one of the worst benchmarks I've seen (even the other "independant" studies Microsoft did over the summer) due to the vagueness of the problem (my 'legacy' is your 'dream machine') to the differences inherant in different operating systems.
As an aside, my 'check' word here is "unguided". How fitting I think.
Want to find other gamers to play board and role playing game
From wikipedia:
:|
"Windows NT 3.1 ran on Intel IA-32 x86, DEC Alpha, and MIPS R4000 processors. Windows NT 3.51 added support for PowerPC processors. Intergraph Corporation ported Windows NT to its Clipper architecture and later SPARC, but neither version was sold to the public. Windows NT 4.0 was the last major release to support Alpha, MIPS, or PowerPC, though development of Windows 2000 for Alpha continued until 1999, when Compaq stopped support for Windows NT on that architecture."
NT 4.0 ran well on my alphastation
Eh. Comparing apples to apples requires PPC architecture. :-)
But seriously, why is it Linux's responsibility to run a gui so you can compare it to an operating system that cannot be pared down to work efficiently in an environment with limited resources?
Ha Ha Ha HaHa HA! Man! That's funny! Yeah. Compare a modern day distro of Linux to ancient versions of Windows. Which, by the way, are no longer supported or being patched. Ha Ha Ha! And then, make a big fuss about how you got M$ DimWoes version Godknowswhat to work on ancient hardware, but the newer distro of Linux doesn't! Oh yeah, this is real NEWS! Ha Ha Ha Ha Ha Ha Ha Ha Haw!
The real funny part of this is that I have Debian Linux 3.1 (2.6 kernel) running right now upstairs on a 233MHz AMD box with 128MB of memory as a server (no gui) and it runs OK. Not the speediest thing in the world, but OK. Acceptable. Try running XP on something like that, and you'll grow old or end up pulling what's left of your hair out!
Besides, older versions of Linux would probably run rings around whatever version of DimWoes that M$ claims to outperform Linux. Yeah. Let's see Windblows 95 outperform RedHat 6.0. Yeah. Good luck with that!
Best Regards,
Kalecomm
It's kind of funny how a 200MHz Pentium, a chip with performance in the same ballpark as a Cray-1 supercomputer, ended up being unusable even as a mere typewriter. I guess that's life in the wacky world of modern software.
The director of the Microsoft linux lab comes right out and says the intention of the testing was to "put to rest the myth that Linux can run on anything.", so you already know that the test has no credibility since its objective was not to find out IF linux can run on anything as is generally assumed. The conclusion to the test came before the test or the results, sounds like standard Microsoft tactics.
Anyhow, reading Hilf's responses in the interview it appears that the tests showed that linux does run on anything based on their test results. He admited that "The tests, which found that Windows performed as well as Linux on legacy hardware" and therefore linux did run on the legacy hardware as installed "out-of-the-box". So the title to the article is wrong as Microsoft's own tests proved that linux would run on the legacy hardware.
Now I suspect that what Hilf wanted to say was that BOTH the Windows and linux installations did not run adequately on the legacy hardware with "out-of-the-box" installs. But he doesn't want to admit it because he actually does realize why there is a wide spread assumption that linux runs well on legacy hardware, because it does.
Note the response to the journalist's question about why there was a "linux runs on anything" assumption, "Hilf said the technical capability to modify Linux, to strip it down to run with a minimal set of services and software so that it could run on all sorts of hardware devices, had generated that larger assumption that any type of Linux distribution could run on all sorts of hardware devices".
And here Hilf is at first correct and then only half correct. It is true that you can strip down linux to make it more efficient and capable on legacy hardware, and it is also true that the latest desktop distros take advantage of the latest hardware and therefore have similar requirements to Windows. But he fails to acknowlege two facts that I suspect he is aware of, 1) even the latest distros can be pared down so they can be efficeintly used on legacy hardware for applications which have reduced resource requirements, and 2) there are light weight linux distros out there which are capable of effectively running on legacy hardware.
Case in point. I can, and have, taken a 533MHz system with a Via Eden processor, 128MB of RAM, dual ethernet cards, and one wireless network card and install the latest Red Hat Enterprise Linux or Fedora Core and have the latest kernel, selinux ACL, iptables, apache, bind, dhcpd, openvpn, and nfs and then proceed to efficiently use the box as a small business web server, file server, firewalled router, wireless access point, caching nameserver, and LAN dhcp server. And the first step is to simply click on only the software packages you need on the box when going through the graphical install.
And the second case in point, as has been pointed out in several other posts I've read, a usable desktop can be made out of legacy hardware using something like knoppix, damn small linux, or any other distro that was designed to use limited resources.
They are really grabbing at straws in their linux lab at Microsoft to try and prove their misconceptions about linux.
burnin
Did I miss it or was the slackware result not mentioned?
From my experience slack is the easiest to put on an old box. Most current versions of Linux cann't seem to handle low res video during the install. I don't know about version 10.1 but 10 installs nicely on 200mhz pentium for use as a firewall.