CIO.com's Esther Schindler has an excellent take on the role that open-source service-oriented-architectures have in enabling enterprise innovation. Sometimes we think of open source as purely a development phenomenon, but it turns out that the licensing behind the code/development is powerful, as well:
The Swedish railroad SJ had a bright idea: Integrate its ticket sales with online auctions. Any seat that hadn't sold by 48 hours before the train was due to leave the station could be placed on the Nordic version of eBay called Tradera.com. Even if the seat was auctioned for a lower-than-retail price, the railroad would get money it wouldn't otherwise see, it would improve train utilization and consumers could get a hefty discount. Great deal. But how do you pull that off with existing IT services? Especially since this project would generate millions of SMS messages?
The answer was JBoss. While still early to the SOA game (which, itself, is still quite early in enterprise IT), open-source JBoss enables enterprises to experiment with SOA in ways that proprietary SOA solutions can't or won't. This isn't unique to open-source SOA - enterprises can generally benefit from open source's "try before you buy, and in bite sizes" - but it's particularly important in an emerging space where enterprises haven't figured out exactly what they want to do.
The enterprise sales model is geared toward big, expensive licenses because it's the only way to justify the cost of a big, expensive sales force. This cuts against the ability for an enterprise to experiment and walk before they run with a given product.
Who do you think is going to win in new markets? Either the massive ecosystem vendor like Oracle or Microsoft that has already bought the customer relationships, or the open-source vendor that doesn't need to buy leads.