I recently made some grades on CouchDB. I found that memory consumption is quite large for building a view (map and reduction), as well as for importing a larger JSON document into CouchDB. I appreciated the view building feature in Ubuntu (4 cores, Intel® Xeon® processor E3-1240 v5 @ 3.50 GHz). Here are the results:
- four hundred 100KB data will cost about 683 MB of memory;
- one set of 80 MB data will cost about 2.5 GB of memory;
- four 80 MB datasets will cost about 10 GB of memory.
It seems that memory consumption is hundreds of times larger than the original JSON dataset. If we use a 1 GB dataset, CouchDB will run out of memory. Does anyone know why memory consumption is so high? Many thanks!
source
share