Hi Couch-istas, Sorry for the dupe question. Thanks for the responses and especially to Paul.
The JSON document model fits my problem nicely, which is my initial attraction to Couch. ...much cleaner than a big table. I also like the idea of using HTTP to plug components together. It's the native protocol of the web, after all. My problem boils down to counting co-occurrence of sets of terms in documents, which can be expressed nicely in terms of Map-reduce. I'd like to distribute the data purely to parallelize and speed up these kinds of queries. The app will serve only a handful of users. My plan for now will be to proceed with a single instance of Couch in the hopes of learning a bit more before diving into distributed Couch. The eventual goal will be to add nodes as we add more data. I'll let you know how it goes. I only vaguely understand the incremental indexing aspect of Couch, and welcome any comments about other differences between Couch's map-reduce and other forms; biting or not. There's lots of cool engineering to explore. Thanks again, -- Chris
