Today we made some tests with one couchbase server and got some odd results.
Stored about 100,000,000 entries with json object (avg 20k size each json)
During the test couchbase went OOM at 50% of the run time and queue to disk was about 10-5MB.
Stored about 100,000,000 entries with gziped json object (avg 2k size each gziped json)
During the test couchbase was healthy.
Why is this difference ? dose couchbase saves the document as is on the disk without any kind of compression ? or having binary/non json document somehow reduce operations that couchbase is doing ?