For years, Microsoft has run a series of commercials, advertisements and infomercials touting the lower TCO of Microsoft products through independent studies. Given the somewhat murky backgrounds of some of these studies, it would seem that a better discussion would be on the exact parameters used for such studies. Alex Fletcher mentions the key cost drivers for a TCO analysis in his blog. I agree with one of his arguments in favor of Open Source - Ease of Integration. In actual fact, integration is one of those crazy ideas that make people like me successful selling services in my regular day job! As a rule, integration is needed when enterprises buy software packages that are not meant to play nice with each other from the get go. Add complexities due to customization and things get even trickier. I am not so sure how SOA is supposed to be the silver bullet for integration woes except that it appears to provide a mechanism to help integrators integrate products. To my mind SOA also leads to cobbled systems, just that the cobbling is a bit more refined. I would much rather configure systems to work together rather than using SOA or some such technology to make integration work. I think Alex's point is well made when he says that integration "can involve business process, policy and governance". Very often, implementation of processes, policies and an overall governance framework can improve integration of systems without costly customizations and SI work.