On Fri, Mar 4, 2016 at 12:05 PM, Jan Lehnardt <[email protected]> wrote:
> It will not :) Darn, now I have to write write a wrapper ;) To answer your question: Since I'm using a shared VMware environment, I ran my benchmark several times, to hopefully mitigate load spikes from other VMs. For each test run, I pushed 3240 documents into couch. 1.6 with defaults took an average of 23 seconds to complete, with a best time of 19 seconds. A single-node 2.0 with delayed_commits enabled took an average of 48 seconds, with a best time of 32 seconds. And just for comparison, 1.6 with delayed_commits turned off took ~115 seconds; best time 111 seconds. Single-node 2.0 with defaults took ~183 seconds; best time 149 seconds. So there definitely is still a noticeable change in write performance - but what I really want to do is speed up view aggregation, so I'm not overly concerned just yet. I'm eager to see numbers for a true cluster, with couch instances on separate hosts, but I haven't had a chance to sit down and figure out the best way to do that with docker (well, actually, in kubernetes is the real goal). Hope that's helpful. peyton
