But if what you say is generally true, then MongoDB is giving a terrible showing for software that's past beta.
I found the following two posts very insightful：
1, http://www.mikealrogers.com/2010/07/mongodb-performance-dura...
"This (mongodb's write plolicy) is kind of like using UDP for data that you care about getting somewhere, it’s theoretically faster but the importance of making sure your data gets somewhere and is accessible is almost always more important."
2, http://www.paperplanes.de/2011/1/10/mongodb_and_data_durabil...
"It's okay to accept trade-offs with whatever database you choose to your own liking. However, in my opinion, the potential of losing all your data when you use kill -9 to stop it should not be one of them, nor should accepting that you always need a slave to achieve any level of durability. The problem is less with the fact that it's MongoDB's current way of doing persistence, it's with people implying that it's a seemingly good choice. I don't accept it as such."
I once tried switch to sharding to avoid high write lock ratio(MongoDB use DB level lock currently, only one write operation allowed at the same time). After sharding, MongoDB even don't know how to count my colleciton. db.mycollection.count() return values at random. mongorestore also failed in sharding setup, there's no error message when I was restoring millions of documents, but after it reported successfully restored, I checked the DB, no documents there.
We switched back to more reliable master/slave setup finally. So before trying sharding, you should watch this video http://nosql.mypopescu.com/post/1016320617/mongodb-is-web-sc... . It's so true to some extend.