Slashdot Mirror
Six Barriers to Open Source Adoption
securitas writes "ZDNet/CNet's Dan Farber describes the six barriers to enterprise open source software adoption. Briefly, the reasons are 1) Lack of formal support, 2) Speed of change (not 'velocity'), 3) Lack of roadmap, 4) Functional gaps, 5) Licensing caveats and 6) ISV endorsements. The article makes an interesting counterpoint to Marc Andreessen's 12 reasons for open source adoption."
server room so as to protect their jobs when they have limited skill sets and no real interest
in learning
anything new.
Code, Hardware, stuff like that.
Newsfollow.com
The IT staff I work for like open source because vulnerabilities for the software we would like to use are found seldomly and when they are found they are fixed quickly, not to mention one program doesn't interfere with the others too much. But management listens to slick salespeople from crappy vendors and we end up with products that won't work with the latest security patches to Windows, and now we are left vulnerable on so many fronts because our proprietary software won't work with the updated and 'secure' versions of Windows. My coworkers could care less about how often the products need updating, as long as staying secure doesn't break our systems, Windows is failing for us in that arena.
Open source at my workplace is stifled by management who don't know the latest tech stuff and listen to vendors more than the folks in the trenches doing the work. Non-tech people are the key roadblock to FOSS adoption, the ever popular 'stupids'.
Even if I knew that tomorrow the world would go to pieces, I would still plant my apple tree. -Martin Luther
Five simple steps for migrating an office to Linux:
1. Build "beefy" Windows 2003 Terminal Server with apps that existing Windows users "have to have"
2. Install favorite Linux distro on all workstations
3. Install rdesktop on all workstations allowing access to legacy Windows apps
4. Wean users to Linux applications at comfortable pace
5. Nix Terminal Server
Ummm... no. Spoke to a novell rep last week (I'm in provo) and they've converted completely (she was complaining that she couldn't use visio anymore).
1. support
Novell, IBM, and HP are teaming up to offer 3 pronged support options (including training) for Linux and the products that Enterprises will run on top of it.
2. installation
... the demo is cool to watch) with the new version of Zenworks from Novell. Installation support is offered by Novell, HP and IBM.
Not only will installation be covered, but so will migrations away from Windows ( automagically
3. deployment
Also covered by Zenworks with its new integration of Ximian's Red Carpet.
5. deploying updates
Again, Red Carpet and Zenworks offer solutions to this.
3) Lack of roadmap
Lack of roadmap, huh. Tell that to mozilla or open office or MySQL or Gnome or perl or
Fedora I could go on, but I think you get the point.
Of course, I prefer a different term than "roadmap" -- vaporware.
SET THE AGENDA?!?! How do you SET THE AGENDA if nobody's paying the bills?
CIO: Hi, we're 2/3 of the way through a $300M 4-year project, and looking for an application that does Foo. I see OpenFoo 0.3 on SourceForge that's pretty close - everything but Feature X with Technology Y - but OpenFoo 0.3 hasn't been updated in about six months.
OpenFoo Leader: Yeah, I got it working as good as I needed it, you can submit your changes to it.
CIO: No, we can't divert resources to OpenFoo - we need to know whan OpenFoo will have Feature X so we can interoperate with Technology Y. When were you planning on implementing it? Feature X would really make our lives easier, 'cuz we've already committed 2 years on Technology Y, and we can't go back.
OpenFoo Leader: Umm, whenever I got around to it. I guess.
CIO: So in the next six to eight weeks?
OpenFoo Leader: Hey man, you want it fixed, you can fix it yourself or whatever! Download the code and figure it out. Who the fuck are you anyways? You wanna use my code, fine, but don't expect, like, a roadmap or anything! I wrote it 'cuz I thought it was an interesting problem to solve. I didn't write the code just so you could use it! Who the fuck are you, man? Like, what makes you think I'm your coding bitch!
CIO: Well, OK, how 'bout we hire you as our coding bitch on short-term contract? $60/hour if you can fix within three weeks?
OpenFoo Leader: Naw, I've already got a day job, don't really have time to dig through all that stuff, and besides, paying for software is wrong.
CIO: Fine, we'll just have to use something other than OpenFoo.
OpenFoo Leader: OK, whatever, man. *click*
CIO: Fuck, we'll just buy 100 licenses from Vendor Z, because Vendor Z has said publicly that they'll have the feature in revision 4.3 that comes out next week, and if not, by 4.4 sometime next quarter, either of which is good enough for us.
Citrix CALs run about $300 per seat. Windows XP Pro is about $300. Kinda defeats the purpose.
Entrepreneur : (noun), French for "unemployed"
does not agree with a specific kernel change?
The author should look closer to SGIs business model. They grappled with this early on and came to the right conclusion:
Linus is in charge, it's his kernel.
Where did this leave SGI, and what does it mean for future development?
They decided:
(1) they can resubmit their changes after adjusting them,
(2) they can provide what they want as an add on (SGI ProPack),
(3) they can forego their project and embrace another one that gets what they want done,
(4) they can choose to not do it.
I was at a conference in 2000, I think, where their head technologist gave a speech on OSS and what it means to SGI. He outlined these options then. I thought about it quite a bit afterword and realized SGI gave quite a complement to the process actually, and Linus in particular. Rather than fight things, they accept them and begin engineering accordingly.
Funny many folks in the audience scoffed at this, thinking the OSS model would get in the way of things. The reality is that is has somewhat, but SGI now has Linux running single image 256 proc machines. Those same machines will run a stock linux, but will run better with the SGI Pro Pack loaded.
Eventually,
(1) Linus will accept SGIs changes,
(2) the kernel will perform the necessary tasks some other way, making those changes moot,
(3) everyone will discover the changes are not needed and move on.
The key here is that users of SGI machines will have clear choices open to them they would not otherwise have.
You can buy SGI IRIX machines that are sweet machines really, but finding applications on them is tough outside of user ported OSS. SGI developed fast and hard early on, but failed to achieve application capture which ultimatly limited their future.
Today IRIX users are dwindling as the mindshare leaves the platform. Make no mistake, IRIX is a sweet OS that can do amazing things, but its closed nature hurt its chance at gaining enough marketshare to survive long-term.
To me, this is a shining example of the primary advantage of OSS over proprietary solutions. Users come first because the process forces the issue, not because it makes more money. Having somebody in charge of core development that is not compensated on its use keeps things clean and workable for everybody.
Linux may not be able to match IRIX today, but the last 3 years or so have been simply amazing really. Give it another 2 or three and it will be there. On a side note, I have invested considerable time and money into IRIX only to see it slowly wither away. Same for various win32 iterations. The primary attraction Linux has for me is that my time and money investments are going to last a good long time. I don't want to go through another transition like that and with Linux, I won't have to.
In the longer term, this kernel is going to eventually spank every last one of the other kernels because it will be developed in a way that actually forwards the art of building and running kernels, not making money. As it continues along this path, the numbers of users grow as does mindshare. You can't get that kind of insurance for proprietary software no matter what you pay because money is the motivator. Think about it, if the software gets too good, what exactly will they charge for?
Clearly SGI sees this because they have embraced the process and appear to be back on track with what they do best; namely, large single image NUMA systems with low latency and high I/O. This time they are building on a kernel that has mindshare and a growing number of applications.
They get to make money, while their users retain choices they would not otherwise have, while they forego the expense of building all that stuff in house. Supercomputing just got one hell of a lot cheaper as a result.
Looking at all of that compared to the proprietary way seems like a no brainer to me really.
Blogging because I can...
NO: MS are obliged to license and can charge for this. No one HAS to license from: if you could reverse engineer before the decision you wtill can, if you could not reverse engineer you can now buy a license. Of course it would ahve been a lot more effective if the EU had imposed free-licensing, but I do not think they get OSS well enough to see the need.