I'm trying to configure a Neo4J instance running on a 4 core machine
with 8 gigs of memory to maximize insert performance via a REST API.
I'm using the Plugins feature to create my own API. I'm also hitting
the server with 84 clients located on separate boxes.
What I'm seeing is that the jvm runs out of memory rather quickly, in
about an hour or two. The only memory related settings I changed are
heap size, and turned on the NUMA option: -Xxmx4096M -XX:+UseNUMA -XX:
What I'm looking for is some insight into what an ideal configuration
would look like for maximum input performance. And any thoughts on
why this might be leaking.