The part of Windows Search that derives from the old Index Server product is actually pretty cool, albeit it feels over-engineered. It has an extensibility API to allow new file types to be indexed. It supports culling search results based on the permissions granted to the indexed resourced for the querying user. It supports indexing other data sources besides filesystems. It has a client/server query API. It seems pretty cool.

It's also frustratingly under-documented and, at least in my experience, suffers from poor performance. The query language is a not-quite-SQL (reminds me of WQL from WMI) and queries return very slowly. I've tried to make analytic tools for file servers using it and when I do get results they're often incomplete and/or wrong.