I've written once on this topic before, but never with Red Hat's assistance. But yesterday I got to present the Red Hat subscription model with two members of Red Hat's legal team - the lovely Jennifer Venable and the ok-looking Rich Bynum (Just kidding, Rich! :-) - at Red Hat Summit. And all throughout the Summit I had open source entrepreneurs asking me for clarification on just how the model works.
First off, it's critical to remember that Red Hat's model works because it innovates. The model requires constant innovation. This is not good if you're trying to milk a product for monopoly rents ("Monopoly" here referring to the limited monopolies afforded through copyrights and patents).
Rich walked through the timeline between Red Hat 2.0 and 5.0, comparing it to Windows's comparable innovation over a much longer period. It is shocking how quickly Linux caught up with Windows and now, in some areas, surpasses it. That is partly due to the beauty of open source software development processes, but it's also partly due to the business models driving innovation in the respective products. And I think it's frankly easier to catch up than it is to do de novo innovation (but the same holds true for Microsoft which played catch up to UNIX for years, and still is in some respects). The real test will be for innovation on a 'going forward' basis.
Microsoft gets paid to rest on its laurels. Red Hat (and the wider Linux world) gets paid to constantly innovate new laurels. Yes, competition from other proprietary software companies can help to nudge copyright/patent holders off their laurels, but not as forcibly. When was the last time that Microsoft Office innovated in dramatic ways? Or Microsoft Exchange/Outlook (or any email, for that matter)? That's competition through IP at its best. Innovate and hold until forced to move. The Red Hat model doesn't allow for such sloth.
In Red Hat's world, much of its innovation (and that created by the wider Linux kernel community) happens first in Fedora, its laboratory. Unlike many commercial open source vendors, Red Hat's intent behind Fedora is actually to push the envelope on innovation, not lag it. The better Fedora is, the more interesting Red Hat Enteprise Linux becomes because RHEL, not Fedora, looks a lot more like the enterprises it serves. More stable and robust. Moves more slowly (but surely). Less risky. Etc.
If you're an enterprise, doesn't this value proposition sound more appealing than proprietary software's value proposition? Big Fees (Upfront license fee + ongoing maintenance and upgrade fees), Lock-in, and Disincentives to innovation.
This "unit of support" is the third critical key, but it doesn't have to be per-server/CPU, the way Red Hat does it, as Rich pointed out. JBoss actually prices support according to CPU Bands (up to 32 CPUs is $x.xx, 33-64 is $x.xx, etc.). For Zimbra, it could be the number of users accessing their email through the Zimbra server.
What Red Hat guards against (and which I can confirm from my own experience) is the customer who says, "We will centralize all support in one person so your support burden is lighter, therefore we'll pay less." Or, "All 1,000 servers will be doing the exact same thing - they're just clustered instances of the application, so the support burden will not truly be comparable to someone else's 1,000 servers."
My answer? Don't get greedy. Red Hat's? Customers should expect to pay for value, and units of service is a good way to measure customer value. (Seriously, if you're an enterprise and paying 1/10th the cost for Pentaho than you do for Cognos, can you really complain with a straight face about how the fee is measured?)
In closing, I'd like to remind my open source compatriots that their primary function is to disrupt. No one wins against a major incumbent by looking/pricing/acting like the incumbent. The road to the customer is through disruption: disruption of the way customers first access the software, disruption in what they actually buy from you, disruption in the innovation a different business model encourages. Proprietary software may encourage innovation, but it does not encourage continuous innovation. Red Hat's model does.
If you're worried that the model doesn't work for applications, databases, etc., let me give you a little data on the results of Alfresco's change to a 100% open source model. We were doing very well in the market before the change, but the move to the GPL has only improved things:
Downloads up 100%
Registered developers up 100%
Leads up 200%
Sales up 150% and on track to exceed 2006 bookings by 400%
Average deal size up 125% (and rising)
- Margins up, too
What, exactly, did we give up by moving to the Red Hat model?
Nothing. Neither will you.