I must own up, I am the Clay Gordon that posted this message, in response to a post on the WWWAC list about the advisability of using Broadvision and a parallel thread on Vignette. I almost used ACS on a project for a magazine group at Time Warner last year before AOL reared its ugly head and before the unfortunate politics that embroiled aD. After doing a cost analysis, I came to a similar conclusion, that using ACS (3.2) over StoryServer/Dynamo (what TW would have required us to use if we hosted things on Pathfinder) would save us over $1 million in the first year of operation, assuming that things were accounted for fully and properly.
The current project, which is being built by OpenForce, involves directory management. We had three groups bidding on the project before I got OF involved. One group wanted 6000 hours of programming time (19 people full time for four months) to do the job. They had no commitment to any platform, but were leaning to using a servlet approach on top of either Oracle or MSSQL7. Another group "only" wanted 2000 hours, but were advocating OpenMarket on top of Weblogic with Oracle for the management system and then publishing static pages to a cache server farm.
The first approach wasn't acceptable because there was no way to assess performance (and therefore the cost of the server infrastructure needed to serve a specified load) as well as the fact that the programmers were in Russia. They were rocket scientists (literally) but I wanted computer scientists because the project was going to be served inside LEO. The second approach was too expensive. Software licenses were prohibitive, programmers were too expensive, development hardware was too expensive, server infrastructure was too expensive, managed hosting costs were too expensive. This is the quote against which I saved over $1 million in the first year ($600k in hosting costs, alone).
I didn't know it would be that much, but I knew, intuitively, that the only way I could build the project on time, and within the budget, was by using an Open Source **system**, not just Open Source tools. Of course, the fixed-price development costing approach was also welcome. None of the other groups would indemnify me against conceptual or implementations mistakes they made.
If people want more details, I can provide them offline or here, whichever is most appropriate.
