Slashdot Mirror
Linux Desktop Myths Examined
Call Me Black Cloud writes "NewsFactor Network has an overview of the $95.00 Gartner report titled, "Myths of Linux on the Desktop". It's a good look at several points from the perspective of a corporate user, not a home user."
... is the flexibility that *nix offers. I would like to see some Win MCSE write a .bat script that could perform half the tasks my bash/perl script foo could handle.
There is still the basic undeniable fact that becuase Windows hides the operating system internals away from the end user, it is far less configurable and less flexible.
This guy totally misses the point on TCO. The thing is if you go with a thin client model -- i.e., have a nice fat server with lots of processing power that can serve up the major appplications to Linux thin client PCs that are, in some part, acting basically as X terminals (although some applications can be seamlessly loaded and executed locally as well depending on demand and needs)
You don't need to spend $BIGNUM on client PCs. Only maybe about $200-$500 a seat in terms of the hardware. And large enterprises don't typically buy their support from Microsoft, they typically buy it from companies like IBM or EDS who then contact Microsoft only when there is a problem they themselves can't figure out. They buy this support whether they have a UNIX client, a Windows client, or a Linux client.... it doesn't matter, the cost of support is basically the same.
This guy really misses the boat, IMHO.
My journal has hot
Is that everyone today is talking about whether or not Linux will really give you cost savings over Windows on the desktop.
A year ago they weren't even ready to admit it was ready for the desktop at all.
Two years ago people would've laughed at you if you even suggested Linux on the desktop for corporate users.
I wonder if next year's report won't be whether or not you should use Linux on the desktop, but rather which distribution you should be using.
The report starts off a section by saying, "Myth: Linux Will Be Less Expensive." The author then shows one situation in which Linux is the same price. "Therefore," implies the article, "it is a myth that Linux will be less expensive." It's an obvious non-sequitur. I wonder if Gartner's clients are paying for that sort of thing, or if it just got added in the summary.
More importantly, the article misses the big difference with Linux, that it puts the customer in the driving seat. If you want to run NT 4 after it is out of support, you won't get security fixes and the like. With Linux, the source code is all out there, so you can keep patching yourself if you want to. Assuming that you aren't running loads of services, that would be a reasonably straightforward thing to do.
This is the reason why Linux is a "paradigm shift" and not just another product which happens to be 10% cheaper.
On a number of points the author dismisses points a "myths" in the header only to allow that they are at least partly true in the body of the text. What he should be saying is that these things are "exaggerations", which isn't the same thing. Calling them "myths" sounds cooler, like he found some big coverup, but it doesn't serve the readers to put up a sensationalist header when all he's really calling for is for the person considering switching to Linux to do their homework.
===== Murphy's Law is recursive. =====
I can only speak from my own experience, but I've been around this business some 15+ years, worked as a programmer, ISP sysadmin and consultant for both really large and really small companies (and a couple of in between ones).
I can't actually recall even one transition from MS/whatever to Linux/*BSD where the people involved wasn't really happy with the move afterwards. They simply never look back.
That's my experience, others may vary, but to me the choice of platform in the overseeable future is very easy. And it's dirt cheap compared to the alternatives too.
The best way to find out is to try it yourself. Don't believe everything you read.
You SO don't get it. Firstly, SMS doesn't work. Everyone knows it. It has failed in every org I've belonged to, and I wasn't the guy running it. People used to laugh about unplugging their PCs so that the IT baboons COULDN'T do an SMS push. ZenWorks might work, only because it's not Microsoft. On Linux, it's all different. Firstly, your "profile" is all in one place. Really. It's called "/home". Secondly, since you can SSH into any box, as an administrative user, you can upgrade whatever you need to. Hell, with a few Perl scripts, you could have the systems autoupgrade - put some .debs or .rpms in a magic directory, and "if -x files, do upgrade process" in a shell script via cron.
The thing about a Unix shop is that you can depend on the fact that all systems have cron, an MTA, Perl, and sshd. In a Unix shop, remote administration is the norm, not the exception.
I want to delete my account but Slashdot doesn't allow it.
The problem isn't security, it's executable content. As long as executable content is never offered in any popular email program (or search-for-ET screensaver) in Linux, we're safe.
actually, the latest version of SuSE ships with executable permission _off_ on any user writable partition. this means that unless the system administrator installed the application system-wide, it can't be run. this almost completely nullifies the virus issue. hopefully other distros will follow SuSE's lead on this point and make this a standard setting on desktop distributions.
"The Gimp ? Yes, and there's Paint with Windows. Not an awesome program but for MOST people it's fine for what they need to do."
Obviously you've never used the GIMP. Does paint have support for:
* Multiple layers
* Layer masks
* Full alpha channel
* RGB and HSV modes
* Advanced Filters
* Numerous, numerous plugins
* The ability to write your own plugins with ~ 15 lines of code
* Numerous brushes, textures, and gradients
* Ability to work with animated images
* Ability to import/export almost every graphic type
I didn't think so.
"Diagramming ? Gee whiz - we all do that all the time !"
Many people do. ORG charts, customer presentations, processes, all require diagramming. At my company, I've had to install Dia on a number of people's Win boxes.
"FTP programs - out of the Windows box too, not that many people need to FTP when they can drag and drop."
IE is not a true FTP program. Real FTP has lots more options. I'm also not talking command-line, I'm talking GUI drag-and-drop.
"CD burning software - XP can write to discs ok thanks."
Can it write VideoCDs? Can it do Disk-At-Once?
"Terminal emulation ? Yet again, how many regular users need it ? But there's Hyperterminal for some emulations if you need them."
Hyperterminal doesn't do 3270. 3270 is a HIGHLY-used terminal emulation. It's used by just about anyone whose got an AS/400 - including churches, banks, department stores, governmental offices, and tons of other places.
"PDF ? Like the Acrobat reader for Windows costs ??" I didn't say it costs. I said it costs time for installation.
"Development software - yet again, who needs that ? We're talking about the desktop here !"
There are many classes of users that use development software. In most companies, the ones who have the know-how can't because licensing is so expensive. On Linux, the tools come with it to do simple GUIs w/ Python to automate tasks. You only have to have one person with a little experience to get leaps and bounds of productivity.
"MS could add loads of extra software in the price but then certain camps would be bleating just like they did with the bundled IE."
See my other post on this topic. Also note that if they could do this within the given price, why don't they lower the price they have now?
"As another poster reports, major corps don't build every machine on it's own ! Wake up - they use Ghost or similar ? At our site it takes TWO minutes to download the build and the scripts add another five minutes to configure the names and IPs, mostly unattended."
Again, read my other posts. Most corporations may use GHOST, but are usually violating a large number of license agreements in the process. That can mean big problems later. Also, to use GHOST, you have to have a _very_ uniform hardware platform.
"And people in-house who can code ? Who's REALLY going to modify their OS ? Version control, in depth knowledge of the source code, testing, documentation, etc. It's just not worth it for most people."
There are many parts that _are_ worth it to people. Depending on what you're modifying, it really doesn't take that much effort. I'd say any company with > 300 people has the tools to do this if they wanted to without much hassle. It's really not any more intrusive than customizing registry entries.
"Linux systems are NOT easier to manage. Read up about Windows GPOs and see how easy it is to apply settings, install software, configure security, etc, to a few or many machines at once."
When I say "manage", I mean update and fix. The need to manage people's desktops in the Windows sense comes from how easy it is to screw up your PC in Windows. With Linux, it's much harder for an end-user to screw up their computer, so you don't need to do as much high-handed "managing". With Linux, you get the ability to do full termina
Engineering and the Ultimate
The real benefits are that money can be saved using Linux if you use Linux within your enterprise for what it is: a network-centric operating system. If you try to simply make Linux work like Windows, you have just forced Linux to ignore its strengths.
The REAL impediments to moving to enterprise-wide Linux implementations are not listed as myths here, because no one ever pretends that these are easy. The big ones:
I don't doubt that these things will eventually happen: Microsoft's continuing increase in obnoxiousness is helping companies along nicely in this regard.
I really believe that one big company, with plenty of internal IT resource, and reason to want Microsoft knocked down a few pegs, could eliminate Windows systems on their own systems (hurdling the obstacles I listed above). This could serve as the benchmark that other companies can point to and see that it is possible. Are you listening, IBM? I'm talking to you!
Look at the tomato! Isn't it sad? He can't dance! Poor tomato!
Let's take two typical software bugs -- one with Windows, one with Mozilla.
Mozilla bug: Submitted bug report, got a preliminary reply via email in under two hours. Bug was solved in two days and pushed back into CVS, ready for compiling. Took under one hour to reproduce the bug, write down all steps in bugzilla, read all the email traffic, and recompile.
Microsoft bug: (registry key not closed on logoff) After waiting 5 hours on hold, I got in touch with somebody asking all the dumbest questions ("Tried rebooting?" etc). The person wasn't even going to give me a phone number if I hadn't asked. I had to be sure to be available at hours when this person would call; I was transferred to three phone support people, and three technicians. I was asked to build two debugging computers and waste a hundred megs of download bandwidth to get certain "debug" patches, only to find that just when I got the computers built and set up, they had managed to solve the problem. Total time spent I working on the problem: at least 40 hours, spread over 6 months. About 10 of these hours were spent answering the same question to new support staff (or sometimes the same staff). Oh, and I was told that I'd have to pay additional support costs if this wasn't a bug in Windows (which it was).
The lesson: "support" is a broad term, and just sticking it on a list of features doesn't mean anything. I'll take the free support from volunteers over Microsoft's any day of the week. Though I have no direct experience with paid support from Linux vendors, I'm confident its quality is higher.
Yeah, we've set up about a dozen Linux servers -- Red Hat and Debian. And there are simply no problems. So the second edge of the "support" buzzword: for the same amount of money, would you rather have support you don't need, or need support you don't have?
These arguments are based on personal experience and not ideals, though I've got plenty based on ideals, too!
Windows 3.1 had code embedded in it that detected if it was running on DR-DOS, and if so, caused Windows to crash and otherwise behave unreliably. (The evidence was presented in court and Microsoft had to pay fines many years later. These amounted to a slap on the wrist for them.) Is there a point to what I'm saying? Yes! A company that puts deliberate bugs into their software in order to crush a competitor might also put special code to detect that an application is Outlook, or Internet Explorer, or Word, or whatever, and show your username next to it as opposed to Admin, just to make people like you feel good. I have no evidence to prove or disprove anything said in my post or in the parent posts. But I'm trying to make a point... Remember the old adage about not believing everything you read? That applies to computer software, too, and probably more so than anywhere else, as people have this way of believing what computers tell them.
Hey, there might be 100 million lines of code in Windows... It might only take 20 or so to put your username next to something that has admin privs.