Slashdot Mirror
GM ponders Linux for 7,500 Dealers
Hedonick writes "Seems like a combination of maturity on Linux behalf and bloat on Microsofts part has prompted
GM to investigate the possibility of replacing their NT based solution with a Linux based for it's dealerships according to ComputerWorld
The whole plan seems to be a bit tentative at the moment but the guy they interviewed, David E. Hutka, is the operations manager for GM:s dealership network and makes some good points on why Linux would be a good (and bad) choice." Unfortunately, it looks like they have a lot of legacy code that may require porting, but it's still good news.
"But a mission-critical application like this really has to work the first time, and there
can't be any conversion issues. "
They've been converting from text based applications you could use from any client provided by any vendor to win95 clients for 5 years now and it's still not done. And if the satellite uplink is down at the end of the month when you are required to transmit Info, the backup is fax and voice.
So it'll take 10 years to turn back.
I think so, Brain. But where will we get a duck and a rubber hose at this time of night? --Pinky
The article is very tentative, the guy hasn't even passed on the idea to his boss let alone received positive feedback on it. Frankly I have a feeling this idea is going to strand somewhere in the higher echelons of GM.
It would be a shame. Linux itself is surely up to the task, even if the support structure has yet to be tested. By thinking this over thoroughly now and doing it properly, in the long run this can save tons of money. You will need to invest massively in training your own people and you will need to secure commitment from the support company that they will be up to the task. There are bound to be a great many non technical users among 7500 car dealerships, you can bet Linuxcare or whichever support they'll get is going to have to hire a lot of people to be able to deal with a customer like that.
It's going to cost a lot of money to do this properly, but not having to pay for 7500 dealerships worth of Win 2000 licenses gives you some spare change to start out with. Of course switching to Linux is a little harder than to another Windows, initially, but eventually the whole thing will be in place and then you start to win. No more paying an OS upgrade license for every employee every 3 years, instead, indefinite upgrades for nothing but the cost of the work of your own IT dept. And then all the other benefits like having the source, reliability, security, etc.
I'd like to know just exactly how "low cost" Linux _is_ in terms of maintenance. I've set up NT and Linux boxes, and every time, it was quicker to prop up... er... set up NT. Of course, flexibility is a minimum, and so on.
But how to they propose having someone with enough of a clue in every dealership to get/keep this running? Office personnel that use computers daily don't even know what directories are anymore. And they expect that car salesman will do this without a problem...?
Face it, NT has more people to whine to, setup geared toward the complete idiot and little dancing paperclips that explain, in monosyllabic words, exactly what you should be doing.
GM might just need a big enough kick to get their asses moving into a constructive direction (i.e. Linux), but I seriously doubt that they'll be getting it from upper management.
Still, it's fantastic that Linux is getting this kind of press-- even if they're only considering it, it makes smaller non-legacy companies look toward it very favourably.
char *death_knell = "If we could change over quickly, if might make sense";
In the long term, pretty much everyone's better of using Linux over NT, and while most people here know it, it's good to see the Fortune One taking an interest.
The point I found interesting is that one of the things that makes this evaluation possible is the fact that GM is going web-based.
<prediction>
this is going to happen more and more as people realise that they need to go the web route, and while they're on the way, they might as well try out an OS that won't cost them anything and won't fall over
</prediction>
I mean, it's nice the guy is the operations manager for their dealer network, but this is really a non-story. Or maybe the story is that a year ago this would have been a huge story, but now, I think it's hohum because I'm only interested when linux gets actually deployed in the Real World[tm].
I really like to see large-scale Linux deployments, like Burlington is doing and GM is pondering. It strengthens the OS by moving it toward critical mass, but I fear it could hurt Linux, also. Companies may now start developing distros with programs added in that were developed in-house and which are not GPL. Ultimately, this could ultimately fragment Linux. Of course, there's no real alternative to advocating the adoption of Linux, it's a great platform and a terrific object lesson of the advantages to free, open-source software.
If more of these big deployments take place, I am in the minority of those who wouldn't be surprised if Microsoft comes out with its own Linux distribution, one that will gradually become incompatible with all the others. Corporate execs who used to believe 'no one ever got fired for buying IBM' now think the same of Microsoft, which is in a position to force proprietary standards (what an oxymoron) on the market. MS may make Linux its next assimilation target.
However, in spite of what we'd like to think, I don't think Linux is scaring the Redmond crowd yet, but that will change quickly if they starting losing chunks of business.
I keep thinking of a penguin waddling up to a sleeping, 800 pound gorilla and nudging it in the stomach. At that point, the gorilla wakes long enough to roll over and smother the bird before falling back asleep.
>Geez... "We don't have enough hard drive space"? ..."We don't have enough memory"? Go buy a couple of 27gig drives for each machine, up them to
;-)
>512 megs of memory.
You're kidding, right? What you're specing out are resources more appropriate for a server than a client. (Then maybe we all need Crays to read our email!
In case you're serious, GM has in the neighborhood of 40,000 dealers across the US. A 27 gig drive would set you back maybe $700 each. 512MB of memory maybe $500. That comes to 48 million dollars; say GM gets a 15% discount for buying volume, it's still more than $40 million.
If someone could walk in with an alternative OS & application, using the same existing hardware & drop a new solution for 4 or 5 million, that's a big win. Careers have been made on stuff like that.
Geoff
I think I see a trend here. Maybe for them it really would be easier to muzzle the entire internet than to produce p
Otherwise, you'd have things like one operating system supporting Unicode only, one system supporting both, and one system supporting Ansi. You'd have to code with special macros and compile for different operating systems! Plus people might accidentally save a file in Unicode and discover that their system doesn't understand it!
Plus, you would have programs that only work on one flavor of windows (I'm sure glad my Hasbro Scrabble game works on Win 95/98 only!). You might even get situations where you buy Windows off-the-shelf and get a different version than what you'd get if you bought a new PC with Windows installed from the very same store.
I feel so cozy knowing that Microsoft is there to prevent stuff like that happening.
Exactly. That's the first thing I noticed too. GM gets their parts from a number of different manufacturers, they should realize they strength of being able to swap parts without being dependent on a single manufacturer. Whether it's spark plugs, or it's an OS. Look at the problems they are going to have now because they used VB.
And although the Linux platform is arguably fragmented, that only affects desktop applications where a company sells a product that could go on dozens of different configurations. Used in a server environment, fragmantation is not a problem. What they need to do is to evaluate a few distributions and determine which one meets their needs best. Say for instance they decide it's Red Hat. Now all they do is make sure that their app continues to run on Red Hat. It doesn't matter if all the other distributions decide to do something completely different because that won't affect how it runs on Red Hat. All they need to be concerned about is what Red Hat does with their distribution. And even if Red Hat changes it a bit so that GM has to make some modifications to keep their app running, I'm sure it won't nearly be the upgrade path that Microsoft has forced on us with DOS -> Windows 3.x -> Windows 95 -> Windows NT -> Windows 2000 -> 64 bit Windows.
Another thing that I hate is the claim that the Server OS platform was fragmented. It was less fragmented the Windows is, IMHO. If I wanted to develop an app for Solaris, it ran fine on Solaris, no matter what AIX was like. However, is my customers wanted to run the app on another platform, which do you think was easier to port my app too? NT or AIX? Which do you think I would consider doing first?
I wrote an essay on the fragmentation in OS's. It might fit well in here.
-Brent--
I've looked hard at porting applications to Linux from Microsoft platforms, mostly VB apps. It's a bitch. There's no VB equivalent for Linux (I know there's some activity on this front), and I'm not aware of an easy porting path (if anyone knows one, comment, please).
There is no question that Linux is a more stable platform than any Microsoft platform. We've been installing Linux machines for our clients whenever we can -- Linux is much easier to work with. However, for GM to convert all those machines over to Linux doesn't make much sense to me. Not only do all of the VB apps have to be rewritten, but also most of the third-party software packages that this guy *thinks* he'd like to add are probably written for Windows (WINE advocates: I can't see my local GM dealer putting up with an "80% working" application under WINE, can you?).
Finally, are this guy's motivations (running out of disk space) silly, or is it just me? He could solve his problem by spending $200 at each store to upgrade to a larger hard drive and maybe to install more memory (DriveCopy does a very nice job of cloning hard drives, small --> large). It would probably cost him another $200 per store to hire a nationwide third-party service organization to visit all his dealerships and perform the upgrade. But at the end of it, he'd have what he needs: working systems with plenty of elbow room to grow.
(Yes, he could take the $400 and probably buy a better PC than he has right now. But I'm assuming he'd rather not IPL them all, install software on them all, burn them in, and so on, which will cost at least another $400 per machine to stage, not including shipping and installation at the dealer.)
If he converts to Linux, he has a huge problem on his hand: not only does he need to rewrite all his applications and do all that expensive staging above, but then at the end of the whole process (assuming it all works) he'll have to explain on a daily basis to his bonehead management how come the XYZ Inventory Control System doesn't run on his machines (his boss spots the XYZ ad in Useless American Cars Weekly, for example).
Ouch.
The VB apps are all on win95 clients with data and communications on the server.
GM's Access project has huge problems on their hands now because it just doesn't work now. They've spent five years rolliing it out and it's still a joke compared to the way the dealers communcated with them before.
I think so, Brain. But where will we get a duck and a rubber hose at this time of night? --Pinky
I work for a group of DaimlerChrysler dealerships, and I find this development *potentially* interesting. For years, Chrysler has had a SCO based server in every dealership. Apart from some application bugginess on their side, the servers are reliable and stable.
However - They went with NT Workstation clients for the MDSII diagnostic system. I suspect a lack of beta testing and a few other items have contributed to continual problems with the things. They went web-based (using Java), and the stability just is not there. The clients crash as a matter of routine, and it took 6 months (here at least) to even get the machines to run the training software without crashing on lesson 3. I would *really* like to see someone give this a try for several reasons:
Longer hardware cycles - without NT crippling the machine, a *nix based client could last longer between upgrade/replacement.
Of course, Linux (*nix in general) has the advantage of being a lot more robust typically. We use a Linux proxy (squid) for web connect and other things, and it basically only gets rebooted with power outages. It's more or less the same with the other Non windows/*nix boxes around here (Irix, SCO, Solaris)
Yes, there are a lot of relatively computer illiterate people out in the dealerships, but I see this as *more* of a reason to go with Linux. If the OS is rock solid, support primarily consists of application issues. That throws a lot of tech support calls out right there.
Finally, to those saying "buy more hard drives" and calculating $250 savings on Windows licenses, don't forget that it takes a lot less server (IMHO) to do the work with Linux. Add to that that Linux is better at multiple tasks, and you may very well wind up saving the cost of extra servers too. Where you might need 3 $10k Win servers to run web, database, e-mail, proxy, blah blah blah...one solid $5k Linux machine will probably do.
Finally, I expect dealerships to become *very* interested in systems that are more open than traditional proprietary solutions. Getting information between the in-house system, the manufacturers system, and the dealership's internal network should be easier...and it can be.
Oops...sorry - that got a bit long winded.
According to a Gartner Group study reported this morning on CNet (http://news.cnet.com/news/0-1003-2 02-114579.html) upgrading from NT4 to WIN2K can be expected to cost a business $2050/machine; Win98-Win2K $3100/machine. At 7500+ machines, that will add up! $15M can do rather a bit of development.
"My opinions are my own, and I've got *lots* of them!"
Last time I was in a GM dealership to get parts was a couple of years ago, but at that time they were using NCD Xterminals. They could bring up all the parts books on the screen, zoom in and out and actually print you a schematic of how to assemlbe something. It was way cool.
The difference between Canada and the USA is that in Canada healthcare is a right and gun ownership is a privilege.
I think that besides the technological advantages to running a free OS such as Linux, there is a more human advantage, which is that with corporation struggling to increase shareholder value, reigning in on their IT costs by not having to pay for Microsoft OS's may save jobs. The money they trim there might have otherwise come from layoffs.
I hope this is something which is actually going to happen, I live near a car plant (ford) and I know many people who have lost jobs due to "restructuring". It would nice to see Gates' billions get trimmed a little due to "restructuring" instead.
five fingers make a fist amalgamate and resist
I think Microsoft knows that once a single IT manager at a Fortune 500 company proves that Linux is far faster/cheaper/better than Windows then it's all down hill for Windows from there, so I'm sure Microsoft is sending the FUD calvary to GM as you read this.
Making the assumption that the Saturn dealerships use the same systems that GM proper employs their problems may go deeper than the OS.
When I was closing the deal on a used Saturn in early 97, I learned their system allows anyone to overwrite an existing record (order). I was incredulous that such a poor design/implementation could exist. If this reflects the total dealer situation within GM, it will take more than Linux to correct its flaws. Moreover, a failure could be attributed to the new OS and not the existing stupidity already present in over abundance.
This hands-on course includes training on Lotus Notes, which replaces DCS as the two-way communication method between GM and GM dealerships.
This suggests to me (I don't know, as I don't work with dealerships) that the applications are Notes-based, not VB-based as stated in the article.
Now, wasn't there an announcement a while ago about IBM/Lotus porting Notes to Linux? That makes this infinitely more do-able.
An interesting point is that those costs are only for workstation upgrades. W2K Server + ActiveDirectory will cost even more. (Although that's probably unquantifiable at this point because nobody really knows how AD works in a large scale environment.)
Business. Numbers. Money. People. Computer World.
What we seem to have here is a mid-level manager evaluating Linux for his company, but he seems to feel sympathy for Linux and be making up the reasons afterwards. It's great he likes Linux, and I hope he will succeed. But I have a hard time believing he will.
Yes, you are right there. -- Another glass of champagne?
It sounds like this is the primary issue for them. If they can easily port their apps, it becomes a no brainer.
I don't know of any easy port path from VB to any of the Linux compilers - presumably a major point in this decision is the work involved.
However, I DO note that they say that their future applications are "web based". If the suit really means Java or CGI, the platform becomes less important.
Unfortunately, this being true, there is more chance they will stick with something like a Windows 9x machine (or keep the existing machines) and go for a large backend app server with thin-client stuff on the dealer's machines.
I am naturally suspicious of this sort of press; it always looks to me like a big company trying to lever more concessions from MS than genuine interest in the platform....
-=DaveHowe=-
...a couple years ago, there were a few products that converted VB to java, C++ or even delphi. Delphi for Linux will be out before long.
They claimed 90% of the VB code goes over, and I saw better results for pure VB code.
Against our flagship VB app, I estimated it would take 6 weeks and three programmers to translate the app over to Java. It had maybe two dozen forms, 3-4 custom components, and several custom objects we wrote.
The pure VB translated fine, but the custom components needed some thought to meld them with the java components, which (in all honesty) were not very mature.
VB has some phat third party controls, and Java can't really match them...as least not back then.
>I mean, it's nice the guy is the operations manager for their dealer network, but this is really a non-story.
Or else he's sending up a trial balloon to see how
a) The car dealer network reacts (e.g., ``No effin' way I want that Linnucks crap runnin' mah computers" or ``Give it a try -- cain't be any worse than that Microcrap on ourn computers"), or
b) See how Microsoft reacts (e.g., ``We can offer you great terms on a nationwide license to Win2000 with the same level of support we offer Intel"), or
c) Making an effort to demonstrate to someone just how hard he's looking at alternatives to their present computer system (e.g., ``I know that it crashes every day. But see how hard I'm studying alternatives? Now do you believe me when I say it's the best system we can buy? And will you authorize another business trip to Acapulco?")
There's undoubtedly more to the story than meets the eye, much of which Slashdotters prolly don't want to know. Let's just focus our attention at making the code better.
Geoff
I think I see a trend here. Maybe for them it really would be easier to muzzle the entire internet than to produce p