Well that's good news, means no changes are needed :-)
I think right now the boosts should come from running parallel view
servers and parallel reduce servers, in parallel. The issue is
currently serialized writes afaik.
Wout.
On Apr 3, 2009, at 8:03 AM, Blagovest Dachev wrote:
This is probably not going to come as a surprise to some of you, but
running couchdb with Google's V8 proved to yield less then exciting
results. In fact I didn't observe any significant speed-up during
view creation.
It actually makes perfect sense when we consider the usage pattern:
many eval() statements with (typically) simple functions, not a
whole lot of loops (if at all), and (possibly) objects with many
different internal classes.
If you are curious, grab the code:
http://github.com/dachev/couchv8/tree/master
