Ironically, SPSS was cloned fairly early on in the OSS wars.
http://www.gnu.org/software/pspp/
I've found that making employees accountable for knowing their software is a huge benefit. Before a number of OSS shifts I've administered, nobody knew what was important. The entire workflow was undocumented. In some ways, tracking down this information is quite valuable in it's own right--and you'd never get it if you couldn't make people's jobs depend on it.
The key is to do it in responsible phases. Pick a representative set of really good people in your workflow. Make them into a "conversion team". Incentivize them to make the conversion process a success. Just doubling existing incentives works really well for sales people. They are notoriously hard to sell on OSS, but 2x-commission brings out the gambler in them. Most importantly--listen to them when they "can't do their work". If you've picked the right people, it'll be due to legitimate concerns.
Go department by department. Be tactical. Allow islands of resistance to form. If they can't be ignored, exploit existing divisions in the company to prevent them from uniting. When they're all that's left in a sea of OSS users, they're easier to deal with. Let their case be about real needs, not "everybody's doing it". Indeed, you don't even have to argue it, their arguments change on their own. It's a remarkably social phenomenon.
The legal department can be your friend. Most organizations are woefully out of compliance in licensing. If legal is made aware of this, they often just can't ignore it and will take it to the top. Ignoring it any any level can make people personally liable. The lawyers will tell them this.
Conversely, if you are in compliance, accounting is your friend. When software licenses are properly budgeted, they show up and they're ugly. It's also fairly easy to demonstrate that, once stabilized, OSS departments require less administrative labor than proprietary ones.
Most importantly, determine where there aren't OSS alternatives. In a big enough organization, you'll invariably have a few MS boxen just for interoperability or niche software. It's fine. That's what virtualization is for, and you can deal with that at your leisure. Rest assured that this is a dwindling list of software.
Be careful. Like any large IT shift, a bad roll-out can negate years of cost savings. No vendor, especially not the OSS community, should be blamed for your botched implementation.
In the end, the dream of an OSS organization is achievable. It can be worth the trouble. Rather you breathe Unix, sleep with a copy of the GPL, hate that your company is probably way out of license compliance, or just want that money in your bank instead of Redmond, there are plenty of reasons to do it.