Clues to the source of some of Sauce Labs' problems can be gleamed from their list of Maintenance headaches:
View indexes are only updated when queried — insertion does not update the index. That means you have to write a script to periodically run all your views, unless you want them to be surprisingly slow when they haven’t been queried in a while. In practice we always preferred view availability to any performance boost obtained by not updating indexes on insertion, but writing reliable scripts to keep view indexes up to date was tricky.
Oh please doing an HTTP GET periodically is tricky ??? No it's not. With couchdb if your database is very dynamic you should either index periodically and very frequently. This creates a quantified and controlled performance demand on the server. Ideally read servers should never be the write server and replication should be filtered. Using Couchdb naively will lead to failure. Don't use javascript views, python views are 3-4 x faster, erlang views are 7-10 faster. Used in the right way and following the many tips that you can get from the Couchdb community will make Couchdb not just a great database but a great application platform.
And yes 1.2 is a great improvement.
Currently I'm using coucdb, mongodb, and MySQL all in one high profile project handling terabytes of data and millions of hits. Each has it's use. When it comes to reliability and performance all three DBs are NOT my problem.
Those who can, do; those who can't, write. Those who can't write work for the Bell Labs Record.