Pixy Mesa compares the Laurel and Hardy of NoSQL databases:
CouchDB is strong precisely where Redis is weak (storing large amounts of rarely-changing but heavily indexed data), and Redis is strong precisely where CouchDB is weak (storing moderate amounts of fast-changing data).
On MongoDB:
Now, MongoDB offers a both a document store and high-performance update-in-place, but its persistence model is fling it at the wall and hope that it sticks, with a recovery log tacked on since 1.7. It's not intrinsically robust, you can't perform backups easily, and its write patterns aren't consumer-SSD-friendly.
Choice Quote:
I do not trust MongoDB with my data.
The comparison table is a good quick-ref of features.
© 2003-2011 Pradeep Gowda