Checklist of things to do to improve couchdb single document operation performance

There are some very simple things that we all can do to get better couchdb performance. I'm interested in aggregating a simple list. For example, I've heard that having sequential id's boost performance incredibly. What other things can we do? What are the most important things that everyone can do, what are your experiences, a开发者_如何学运维nd what has helped you most?

Have a look at the Definitive Guide: http://guide.couchdb.org/draft/performance.html

Some other points:

• sequential UUIDs instead of totally random
• don't use include_docs=true but emit full docs (view size <=> speed trade-off)
• prefer batch and bulk operations