I also don’t think you’ve taken into account when happens when one needs to “scale” mysql past the confines of on server. You can add as much memory or CPU to a mysql box as you want but, in an optimal environment, the limitation will almost always disk I/O. An IT, developer, or otherwise thinking they can just “move to RDS” to solve their mysql scalability issues should rethink their situation — Amazon RDS garners little gain in I/O over a dedicated server. Except for small shops, without extended support for replication or other beneficial features, there is simply no benefit to migrate to RDS over simply understanding how to properly configure and deploy mysql within your own environment.

If my code are reusable (now that RDS is there), why won’t I move my apps to cloud and try get rid of my IT overheads (and the headache) ? Didn’t we outsource to India and China and focus on our core operations ? instead of being at the mercy of IT dept in US ? Thinks it as IT function outsourcing, while keeping the IS function in house (same as splitting design and Silicon fabrication functions in 80′s in Chip business)

So from Amazon’s (and all our) viewpoint it makes sense -
1) It lowers the entry barrier to users and developers to get into cloud.
2) It enables faster adoption of cloud, and thereby brings in investment to further stimulate cloud tech development (remember Ford had said -”..they can have any color as they want as long as it is black…” about model T, whose standardization had helped build economy of scale and helped the auto industry)

The argument against RDS would have been stronger if the above analysis had done a comparison on economics and argued from business and strategic angles, rather than speculating that my own backup would be much better than Amazon’s professionally managed data centers.

Sure RDS is relational db thinking, which is not the cutting edge, but it still has its role to play (assuming the RDS is much faster that SimpleDB, else RDS and SimpleDB has to compete with each other) Amazon should be careful against such possible internal conflict and canabalisation, and the RDS engineering team have to keep better latencies in RDS compared to SimpleDB, would be interesting to see how each team does that over next 3 years).
I am sure Amazon’s executive may have thought about such Darwinian strategy to stimulate tech development internally.