the major problem with these "studies"...
|
| Author | Content |
|---|---|
| tuxchick Nov 18, 2005 9:54 AM EDT |
...is they contradict the experiences of real-life system and network administrators. Why would anyone go through the pain of migrating Windows servers to Linux servers? (Or BSD, or Unix, or what-have-you.) It's not that we're bored and think it's a keen idea. The SUSE problems in the report sound bogus. As you said, any sysadmin worth his or her stuffed plush Tux knows how to handle a glibc upgrade, and what to do if it fails. And if they did have problems, why didn't they call Novell for help? Isn't that why you spend money for a vendor-supported distro? "Security patches were applied in one month increments." Sure, why not give the black hats a window of opportunity. But what I'm really wondering, is what were they doing? What applications were they running, for what kinds of jobs? Under what conditions, on what hardware? Maybe SEL is poo, though I doubt it. Maybe they're blaming SUSE for problems caused by the sekkrit 3rd-party software. Anyway, 'funded by Microsoft' is pretty much all you need to know. |
| tadelste Nov 18, 2005 10:10 AM EDT |
This study verges on Consumer Deception and Consumer Fraud. http://www.usdoj.gov/fraud.htm |
| number6x Nov 18, 2005 10:52 AM EDT |
"Consumer Deception and Consumer Fraud" Tom, That's just a nice way of saying "Marketing". :) The study may have been accurate, but remember that when you pay for the study, you get to specify the parameters. Like the cost comparison study Microsoft payed for comparing having your mail server running on Linux vs. using Exchange on Microsoft. The study showed Linux costing hundreds of times more than Microsoft. The study was accurate. They were comparing the Commercial Bynari server running in a mainframe LPAR on Linux/390 to Exchange running on MS Server 2XXX on Intel. This study might be "truthfull", but the parameters are set from the start by the party that commissions the study. The "analyst" group just follows the parameters, a little "shake and bake" in the testing labs, and Voila! "Microsoft is better than Linux in task ." No one has to lie or deceive, just set the tests up "correctly". |
| hkwint Nov 18, 2005 11:00 AM EDT |
Quoting:Maybe SEL is poo, though I doubt it. I think we can find the reason why they chose SLES on page 27, last paragraph, sixth line: "At the time of SLES 8 release, all three components (AMP) were behind the most current versions..." So they knew, it would require more updating then RedHat. This is the first time they don't pit MS against RedHat, so its suspicious. Having worked on the LXer migration list, and heaving read a bunch of the 'succes stories' (Novel adds) and other Suse-stories, I think I can say, reliability is Suse's main selling point advertised by Novell, so they hit Suse where it hurts most. |
| tadelste Nov 18, 2005 11:13 AM EDT |
Well, since I founded of Bynari and invented all the products, I think I can speak authoratatively on the issue. But, I won't say anything else about Bynari. I will say that Linux doesn't run as fast on a S/390 as a zSeries. It runs exponentially faster on an xSeries (Intel) box than an iSeries, pSeries or a zSeries. I also took part in a lot of benchmarks of Bynari on IBM S/390's at IBM and depending on the benchmarking software, Insight server generally slows to about the speed of a 486 intel machine. Change the benchmarking software and make a couple of configuration changes on Insight Server and it runs like a bat out of h*ll. Not marketing: making the results say what you want them to say. I can do that too! |
| number6x Nov 18, 2005 11:34 AM EDT |
tadelste, I think that that comparison was on cost alone, throwing the mainframe in the mix made the "Linux" mail server cost much more than the Windows. Bynari and Linux cost less, it was the hardware specs that skewed the results. "New study shows that Windows Vista and a pack of gum cost much less than Linux and the Wrigley Building." |
| tadelste Nov 18, 2005 11:36 AM EDT |
number6x: I understand. We used to get better performance on a $4000 xSeries than a $1million S/390. |
| GrueMaster Nov 20, 2005 7:19 AM EDT |
After thoroughly reading the study, I was really impressed with the fact that the Windows users recruited had significantly more experience than the Linux users. I also found it interesting that they started with an unpatched version of Linux vs Windows 2000 SP4 (signifying that Windows had already gone through significant patching). I also noted that when they installed a major component on Linux that required newer versions of various applications, they decided to stray from the distribution supported updates instead of just doing the release upgrade at that time. If they had upgraded at the beginning when called for, they more than likely would not have experienced dependency issues, and the fact that they couldn't reboot after 10 hours really shows the lack of skills in the Linux administrators. While I agree that supportability is a major factor in upgrading a server OS, New feature requirements is also a major factor. This area is where Windows really lacks. There aren't very many new features between W2k & W2k3, just more bandaids and a pretty wrapper. You can't heal a wound just by changing from a Snoopy bandaid to a Scooby-Doo bandaid. Without proper care, the underlying wound just festers, waiting for an infection. Also of note, none of the Linux patches were specified, making me wonder what patches were required. If they were specific to vulnerabilities, they should have been listed. If they were things like "Vim causes a buffer overflow when editing piped input from /proc/kcore, causing xmms to corrupt an id3 tag in the song currently playing", who cares? (note, this is a fictional bug, although I have seen similarly obscure bugs being reported). I recently upgraded my heavily modified Mandrake 10.1 desktop/server system to Mandriva 2006. Yes, I did run into some issues with components I had that were outside the distribution, but I planned for those before hand by installing them outside the normal distribution path (/usr/local). That way, I could figure out what components were broken, and what external components were integrated with the distribution and no longer needed. And if they really needed to upgrade glibc, why not install it locally to the application that needed it? Not once in the study did it really go into performance, or reliability, areas that Linux really shines. I almost wonder if they even bothered using identical hardware. As far as 3rd party developers using later rev's to develop their products, they usually do that due to major innovations in the newer software (MySQL 3.23 -> MySQL 4.1 had major feature changes to the core, increasing reliability and performance). I have been a Linux user since 1996, and started doing Linux administration in 1999. I am completely self taught (go O'Reilly!), and have successfully rebuilt a CPU optimized distribution from source (Redhat 7.1), deployed an Apache/MySQL/PHP server, developed a few kernel drivers for in-house proprietary test equipment, and deployed a test farm of ~250 systems spread across 3 countries globally (before my job was outsourced - gotta love the corporate mentality). |
| Bob_Robertson Nov 20, 2005 12:24 PM EDT |
When mentioning "patch" volumes, the idea that there are substantially more patches issued for SuSE than for Windows, did they mention if they were patches only for core "Windows" compared to all of SuSE announced security updates for several thousand different applications? Just a thought. |
| bdumm Nov 20, 2005 12:41 PM EDT |
Snoopy? Scooby Doo? Man GrueMaster you are showing your age.... ;) |
| hkwint Nov 20, 2005 1:32 PM EDT |
Please read the MS comments defending the study: ( http://www.eweek.com/category2/0,1874,1237915,00.asp ,3d page ) Ryan Gavin, Microsoft's Director of Platform Strategy, defended the need for the study, even if it was just a starting point. "There is a big need for a larger understanding of how we think and talk about reliability and to have a consistent dialogue about what things matter when you are talking about a reliable system," he said. "One of the reasons we worked with Security Innovations is because they are more academic and research-oriented and we wanted a repeatable methodology that could be applied again and again," he said. We should send this Mr Gavin an (open) letter, and ask him about his ideas what he means with academic and repeatable, and the distortion of the facts in the press release. If anyone can find or knows his email, please let me know. |
| tuxchick Nov 20, 2005 1:50 PM EDT |
hkwint, they're so full of it having any sort of a meaningful discussion is impossible. How do you know when they're lying? Their lips are moving... ;) But an open letter would be fun, and a good vehical for making factual points. Whatever response they make is good for entertainment value, and not much else. |
| tadelste Nov 20, 2005 3:12 PM EDT |
Quoting:"There is a big need for a larger understanding of how we think and talk about reliability and to have a consistent dialogue about what things matter when you are talking about a reliable system," he said. This clinically pure and 100% arrogance. If you read this correctly, this guy is saying that "we" need to understand how "they" think and talk about things. That's someone who speaks as if he's entitled. I would reply to this person: We don't need to understand how you talk and think. We would rather you take a spaceship to another planet and set up shop and walk around slapping each other on the back. I don't like your work and you have no right to demand I join it. You know you have no security and you've been hiding the fact for years. You allow people with no technical understanding what so ever to surf the Internet with Netbios ports wide open and allow people to install rogue programs on others computers through your browsersn and email products. Microsoft's response is just double-talk. They've sold their souls for a few temporal dollars. |
| helios Nov 20, 2005 4:57 PM EDT |
I have nothing of value to add to this discussion, simply an observation: "New study shows that Windows Vista and a pack of gum cost much less than Linux and the Wrigley Building." You don't get the in-your-face perspective like this anywhere else but LXer. Number6x slammed the door on the comparison with 20 words. helios |
| hkwint Nov 21, 2005 4:03 AM EDT |
Quoting:Whatever response they make is good for entertainment value only. Yes, you're right. When we need a funny story again, we should think of this lad. |
Posting in this forum is limited to members of the group: [ForumMods, SITEADMINS, MEMBERS.]
Becoming a member of LXer is easy and free. Join Us!