Alberta Heart Institute, CMS complaints

The Alberta Heart Institute web site went live on Friday. This was the first time I've used Microsoft Content Management Server to build a site from scratch and it was, for the most part, okay. Michael Warf provided the design, and did some simple-yet-cool Flash components that integrate with MSCMS. There are still many improvements to make, but the clients are happy.

One of the challenges I've had working with MSCMS is understanding that content is stored as objects and not as records in a relational database. So, apparently, you can't just write a query that grabs all the pages with the keyword "arrhythmia." You have to recursively crawl through the whole site tree (i.e., every page) to find that information.

This seems like a staggeringly inefficient way of finding related pieces of content, especially ones that aren't on the current branch. Which makes me wonder: If this kind of object-oriented content model is so efficient at presenting traditional hierarchies, is it at the expense of valuable contextual connections across branches of a hierarchy? It's like my content is chained to the taxonomy, and I can't leverage my metadata to set it free. Or something.

One solution I thought of was to crawl sections of the site nightly and dump the metadata and url of each page into a database. But a sophisticated CMS shouldn't force you to resort to such a kluge.

If you know a way around this problem, please enlighten me either in the comments or by email.