Slashdot Mirror
Stories of Open Source Failures?
ahodgkinson asks: "We often hear about companies, government agencies, schools and other organizations that migrate from Microsoft to open source based systems. We sometimes hear about organizations that evaluate Open Source and then elect to remain with their existing proprietary system. Both of these events represent represent a 'non-failure'
for the open source movement. I'm interested in knowing more about the Open Source 'failure' events, namely when organizations move away from open source to a proprietary solution. Does anyone know of organizations that have moved from an Open Source based IT solution (back) to a proprietary system? Or where such a move was contemplated but not made? I'm specifically interested in larger organizations that have 'undone' a strategic move to Open Source, and their reasons why. Given your examples, is there anything we can learn from them?"
...because I'd wager that in most cases, people choosing to deploy open source solutions are driven, and do not accept failure. There have been plenty of times where I could have allowed an open source solution to fail, but persevered and eventually made it work the way I wanted. So while I've had plenty of setbacks, I've had precious few actual failures, if any.
It's still rather early in the game for there to be many failures. Though the momentum is growing, there are still very few businesses that have made the jump to Open Source. With only few businesses trying it there are only few chances for failures.
I'm sure that there will be failures. There are always failures, even in proprietary software shops. There are many major IT projects that have been based on well known and respected proprietary applications like SAP or CA Unicenter an a slew of others that have failed miserably.
The failures will be due to many factors, poor planning, poor implementation, poor software or who knows what else. There will always be failures and as Open Source spreads into enterprises around the world there will be IT projects based on Open Source that will be abysmal failures.
But, the fact that there will be failures doesn't mean that the concept is a failure or even that the software is a failure. As I said there have been many multi-million dollar failures with the likes of SAP and CA but, I don't think that anyone would classify either of these companies or their products as failures.
One last note: If you are looking for failure, you will surely find it. Why are you looking for failure?
Well, there's Microsoft switching from FreeBSD to Windows. But does that really count as a failure of open source? After all, other than during the many failed attempts at transitioning to Windows, Hotmail ran extremely well. And the cost factor is rather skewed when you get as many copies of Windows as you need for free. And it was corporate pride/image, not technology, that drove the change. Still, they did finally change. ('course then they blew up their DNS and ended up outsourcing to someone running it on *nix so I guess there is balance in the universe.)
~~~~~~~
"You are not remembered for doing what is expected of you." - Atul Chitnis
On one project, we used PostGreSQL's GIS extensions, PostGIS under SuSE for the prototype, as the prior GIS DP methodology was to do all the GIS processing by hand on a windows desktop--which read and wrote .shp files. Gross! After developing a prototype
DP stream in PostGIS, which is OGC compliant,
it was fairly simple to migrate the DP methods (all SQL with OGC-compliant GIS data formats
and stored procedures) to DB2 Spatial for the bulk processing, which could handle even
larger data volumes, and much, much faster. By about an order of magnitude.
Hours instead of days. Is it an "open source failure" to prototype a process using
an open source tool, then migrate it to a proprietary product that's actually better?
Both still ran under SuSE.
It demonstrated the utility of doing the GIS processing required with a spatial
database rather than a silly little pointy-clicky windows app. Without the OGC standard
that both PostGIS and DB2 Spatial adhered to,
however, it would have been a real nightmare.
Which were not failures of the FOSS per se, but represented usually the success of an incumbent with a strong preference for Windows winning a political victory, and in a few cases the FOSS exposing existing problems and being blamed for those problems.
For an example of the former, consider a client that owed me AUD$3000.00 when they went bankrupt from accumulated incompetence. They had a Linux system replacing a Novell box (and incidentally taking a load off several Windows boxes) at a site with a variety of Windows (3.1, 3.11, 95, 95C, 98, NT4) with a 13GB x 2 RAID1 array and a UPS - and random networking issues which appeared to be in the wires since we replaced everything else and it still went funny every so often.
Aaaanyway, an incumbent manager had achieved "golden boy" status by signing up a contract on the other side of Australia which was a complete shoe-in (my cat could have done it, just tuck agreement and pen under collar, put in pet crate, address and ship) and really liked Windows.
He ran their time clock on his own Win98 machine and refused to acknowledge that this was an idiotic thing to do, even after many times losing most or all of a morning's time records because the machine had crashed before or during the arrival of their workers. He eventually would up shutting the machine down at night and having the BIOS wake it up at 4:30AM, thus cutting his data losses down to oince amonth or so... I'm sure you get the idea.
Mr Golden Boy had arranged to get me kicked out of the place, bills unpaid, on a Monday and that Thursday they had a power failure. Shortly afterwards, one of their staff walked past the server room and noticed a buzzer sounding and a light, so being the helpful little sod that they were, they switched off the offending device - the server's UPS.
When the power came back but not network services, somebody else figured out what had happened, and switched the UPS back on. After ten minutes, still no joy, so they called me in. Not Mr Golden Boy, not the uberManager, they got one of the few remaining staff with a clue, one of three in the place that I cared about, and got her to ring and plead for them. Scum!
I drove half an hour to get to the place, looked at the server and it was mid-fsck (13GB software RAID one, old machine, you get the picture). As I left the server room I met Mr uberManager, who asked what was going on. I told him that the machine had been repairing itself after being interrupted and that it was taking a long time because of the large hard disk capacity, probably twenty minutes to go and it would fix itself. Mr uberManager nodded, turned away, and I turned around - to find Mr Golden Boy looking like Zeus on a bad day, red-angry and fit to apoplexy because their company's server and all of its data were going to be OK! What chance did I or FOSS stand in the face of an attitude like that? Hint: it comes between "9/(" and "-/_" on your keyboard.
For an example of the latter, consider the first round of StarOffice Wars some years ago, where they lawyers in question had sucky/random document structure and had to pay the ferryman anyway when their old Kyocera printer died and the new one had slightly different layout.
In summary, you will get different answers depending on how people percieve your question. I predict that there will be many political failures, and a very few FOSS failures reported.
Got time? Spend some of it coding or testing
Slashcode?
"Eve of Destruction", it's not just for old hippies anymore...
We do, don't we? I'd actually like to hear some follow ups on these stories that are always promptly reported as a victory of sorts.
For example, how long it took to actually migrate x,000 of servers and workstations after the [government | company | school] decided to "give M$ the shaft". How much money for re-training users? How much lost (or gained) productivity? How much churn on the HR side because admins|programmers could not cope with the platform change? How much cost for replacing or rewriting business applications? Buying new ones? And so on.
I've always thought in looking at those "success stories" that they were rather long on hype and short on substance. I personally know of a few successful moves to things like OpenOffice or different mail servers and databases, but never a wholesale large scale Windows->Linux migration that in the end actually worked to everyone's satisfaction and ended up being cheaper than it was before.
I set up a network for a previous client, a large private middle school, based around FreeBSD/Apache/MySQL/BIND. It was a nice implementation; very secure, utterly reliable and as much open source as I could get in there (I was not able to move some servers because of Windows apps the previous guy had installed on a few boxes that they just *had* to have.
Our initial plan for this client was to move everything to Linux-based Xterminals (after all, what do they do? Edit some word docs, look at web pages, send mail) but management decided that the time wasted by some clueless idiot coming in after we'd left who didn't know what he was doing would outweigh the cost savings.
So, I snuck FreeBSD in as their monitoring, web, DNS, and firewall server, not to mention software repository, UPS controller, and groupware server, along with a host of other functions.
I still check up on their infrastructure occasoinally, and have noticed that the guy who took over their support after we'd left has been steadily moving everything back to Windows 2000 as fast as he can; he runs a small IT "consultancy" and just can't be bothered to learn how to use something that doesn't require point-and-click.
Regardless, I consider it a minor victory that some of the services I set up (firewall, monitoring, etc.) have withstood any attempts to downgrade them to M$ brokenware--if only because nobody could figure out a way to do it better and easier....
Sigh. Oh well, they paid their invoices on time.
Cole's Law: Thinly sliced cabbage
expecting the machine to handle internal people mailing 10 meg+ attachments to 900 people at once and not buckle under the load
Incidentally, Exchange handles this easily - it'll store one copy of the attachment and just put a reference to it in 900 mailboxes.
Microsoft's Exchange marketing spiel (shared folders! forms! scheduling!
If you need these things, you'll need Exchange or Notes. Open source simply doesn't have those features. Sure, you could probably implement them using Open Source (i.e. writing Perl CGI scripts) but why would you?
This really isn't a failure of opensource. It is a failure of the insurance industry, and specifically your company to come up to the plate and take care of the situation. Open source is not your free software repository which you can take from without giving back. Hire a competent coder to look into doing an opensource of TAM.
Not only does the general economy improve by having an employee that has ability to spend money, but you can sleep well at night knowing that you have also given something back to the opensource community and the insurance industry.
Just because opensource coders who are doing this on their own time are working on what they consider to be cool things, doens't mean that it is opensources fault that you don't take the initiative to correct the problem. Hence the term community.