Slashdot Mirror
Linux & the Business Desktop
Ulwarth writes: "Desktop Linux is running a feature documenting a mid-sized company switching to Linux on the desktop, like the City of Largo but this time in a corporate environment. Proof that it can be done - at least for businesses which need only the 'standard' office apps."
I know that Linux has some decent GUI's available, but I would wager that the final push needed to get it more mainstream is for someone to make a GUI that looks 95% like Windows GUI.
Don't complain, don't tell me that Linux GUI's are better. That's a moot point. We're talking about end users who just want to sit down and work with a minimal amount of retraining and confusion.
IT departments can be as smart and savvy as they want to be, but in the end it comes down to simplicity for the end user. Make that and Linux will have a much better chance. When no one notices that they're using Linux, you have succeeded.
------
Today's Top Deals
i work at one of canada's banks, at the corporate university. another comp sci guy and i were successful in moving most of our servers (~25) from NT to linux or back to solaris (bank *loves* solaris).
now we're attempting to deploy linux on 20% of the desktops (~10 people).
we've already realized that we'll need to cram VMware on some of them (flash developers) but i consider this just a transition period. hopefully things will go well and we can divert some upkeep dollars to R&D...
- Applications
- Training
- Accountability
Linux will never become big on the desktop until something is done about these major issues. Linus, Alan, et. al. need to get off their behinds and tackle the major unresolved issues behind their OS. GUI developers need to quit worrying about transparent windows and drop-shadows and get working on making their systems stable and functional.When a business upgrades its systems, it wants proven reliability in its equipment. Applications such as StarOffice and KOffice are high on glitz and glamour, but lack the backend to fulfil this stability requirement. While Corel and Microsoft focus on ensuring their system works before incorporating new "features," the uncoordinated Linux effort works to force nifty "features" onto an unstable backend. This means that while I can do some interesting graphic and font modification, my attempts to save throw Kernel panics and crash my system.
Almost 100% of office workers in the present work environment have been trained to use Microsoft Office. Most students come out of college having used Microsoft Windows as their OS, Microsoft Word to type their papers, and Microsoft Excel to do math projects.
Switching from a Microsoft base to a Linux base means a great deal of downtime while workers are retrained to use their new desktop environment. System Administrators must be trained or hired to work with a new system base. Technical Support people must learn how to handle the millions of innanely obtuse error messages thrown by any one of 1000 different applications installed by default on the new systems. (Why is xterm crashing with a tcpdump error message? I'm not running tcpdump!) Each user must be trained in how to login to their system, navigate a new and dramatically different desktop, then they have to be trained in how to use a brand new office suite. While this process can be spread out using staged upgrades, the downtime still adds up.
In the end, the Linux kernel is maintained by a group of hobbyists. As with the applications, these hobbyists put a large amount of time into programming glitz and glamour features into the kernel, and neglect important functions such as scalable SMP support, efficient VM managment, clean TCP/IP communications, and such. These important functions end up being "fixed" by other hobbyist programmers whose fixes usually end up making systems less stable.
While there are groups available for support, many of those groups are closing shop because they aren't getting business. It's a vicious cycle. Anyhow, there's only so much those support people can do. (Red Hat: "We can't support that because none of our people have used it or tested it with Linux. Look it up.")
When Linux becomes more like Windows, more people will use Linux. That is a fact!
When I became a man, I put childish ways behind me.
I am watching this trend with interest; last year there was one customer like this per month; so far this last 3 months I see one or two per week. Much like the ramp of Thin Client or web based computing this trend hopefully shows that critical mass is forming.
Evil ZEN Scientist
pronoblem
2. The vast majority of custom, in-house developed apps will never be ported to Linux. (It's just too expensive to do the massive rewrite needed to port any non-trivial program written in VB or VC++ to Linux.)
3. Apps like (1) and (2) above are "must have" items in the vast majority of mainstream (home and office) desktops.
4. Because of (4), Linux + a GUI interface could be 1000 times better than Windows, but it still would never be more than a niche of a niche on the mainstream desktop. No level of security or usability or cost savings or any other benefit will ever overcome the fact that if it doesn't meet people's needs it's useless. And no amount of wishful thinking will make it otherwise.
(Don't flame me. I'm a writer and programmer in the Linux field who's invested a lot over the last few years in trying to help Linux succeed in the mainstream, and I'm deeply frustrated that it hasn't done better. But I've also talked to many consulting clients in companies of all sizes about conversion issues, and I'm convinced that the above 4 points accurately reflect the situation.)
Find me a story where a company that has a $100M invested into their custom accounting/billing solution has decided to throw it out and spend another $100M to rewrite the software for Linux.
Where I work, custom software is the norm. We purchase almost nothing in terms of off the shelf packages and the main system just does it all. The code has been around for about 20 years or so, ported from platform to platform. Currently we're on the NT platform, but after a few bullying letter from MS stating that since we haven't purchased anything in the last year or two from them that we must be pirating software.
Well, that was it, we started looking for a way to move off the NT platform. Our codebase is roughly 4.5 million lines of COBOL code, and the data is archived back many years as well. So we had 2 problems, there was no COBOL compiler for Linux, and anything resembling one was not data-compatible. Until about 4-5 months ago that is, now MicroFocus has spun off from Merant and become a single entity (again). So now there is a source and data compatible compiler for our backend apps. Just change around some directory separators and it just works. No export/import/etc.
I guess my point here is that I've found most anything can be targeted for the Linux platform, especially those systems that are character-based, as many that I've seen are. The GUI interfaces are relative newcomers to the field. And with borland porting toolchains to linux, I'm sure it will just get easier to retarget -> recompile and have shiny new binaries. Even better when the toolchain is just a port by the tool vender, then most likely your data will be fine too. The only exceptions I see that will probably never change are the MS toolchains, MSVC++, VB, VFP, etc. And if you're starting MS, then you probably don't have much choice of platform, though I have seen a gnome project somewhere trying to build up to source-level compatibility with VB.
And I don't want to here any whining about "what if you don't have the source," all I have to say is, if you payed $100M for some software and didn't even license the source, you've got bigger problems than blue screens and crashed desktops. We are a medium sized company, and our code didn't cost that much. And we get to keep it to if the developers go under, plus add anything we want to the source. Now that I'm somewhere that has done that, it makes no sense to buy software for XX thousands or millions of dollars and not get the source... That is a huge risk.
Free Online Woodworking Resources Directory
Believe me, when Windows 3.0 was being pirated like mad, creating a market for those old Excel versions, there was no training or accountability. People were used to DOS and keyboards; and what "accountability" was there for pirated software? Once a "killer app" came along, it took off.
The arguments you use here are excellent arguments for why Linux can and does do well as a server. However, they don't help Edna the Accountant out very much.
Remember, Edna is not a power user. Edna is not used to reconfiguring IIS, or Apache, or O'Reilly Website, or sendmail, or Exchange based on profiles, subnet masks, header information, etc, etc, etc. System Administrators are an interesting lot, and they love the challenge of learning and tweaking a new system.
Edna, on the other hand, doesn't care whether Exchange or Sendmail is handling her e-mail, so long as her happy forward about Joey and his paintbrush makes it to her 3 daughters in Illinois. Edna, however, is concerned by the fact that her start button has been replaced by some strange set of icons along the bottom of her screen, and the fact that her cell-formulas for calculating the interest-rate based on monthly payment have changed.
Edna doesn't want to be bothered with learning how to run XSpreadEmSheet or whatever. For her, Excel works perfectly and will continue to work perfectly. Getting Edna to change is like getting a donkey to walk down stairs. You can do it, but neither you nor the donkey will be happy doing it.
Another flaw in your argument is that you automatically assume that the difference between Windows an Linux is as pronounced as the difference between DOS and Windows. Back during that time, Windows was a very big deal. The only GUI available had been for Macs. For the first time you were able to navigate your computer without remembering long confusing paths and command line options (Server admins love command lines, Edna hates them more than vampires hate garlic). Today though, most people are plenty satisfied with what Windows and MS Office can do for them. There is very little incentive for them to change that. To them, Linux is no "killer app" (A point you do concede).
Linux doesn't, and in my opinion, will never have a "killer app". However, if you can use Linux and KDE on a desktop for basic office functions now and save $400 per computer, that is really the "killer" aspect of it.
Let's look at this from a business standpoint:
Windows and Office for each computer: $400.
Training/break in time: $1000
Windows Total: $1400 per user
Linux and StarOffice for each computer: $50 (Media and such)
Support for Linux and StarOffice: $100
Training for old and new employees: $2000
Lost productivity as old employees learn new system: $5000
Linux Total: $2150 - $7150 per user
This is what companies see. Yes, Windows "costs more" ($50 vs $400) per computer, but Linux "costs more" per user ($1400 vs $2150 - $7150).
I'm the tasty treat nobody can resist!
IM Me! AOL IM:Tasty Beef Jerky