I remember reading that when the developer of Sphinx was benchmarking MySQL's fulltext searches at Craigslist, most of the time spent performing a query was spe

I remember reading that when the developer of Sphinx was benchmarking MySQL's fulltext searches at Craigslist, most of the time spent performing a query was spent in locks and mutexes. The actual query time was very fast, but the overhead was what killed performance. From what I understand, Postgres doesn't (necessarily) have those kinds of locking issues, but MongoDB does let you fetch documents (especially hierarchies) in a much more simple manner, rather than fetching them via potentially complicated join queries.