I'm concerned with the amount of CPU Neo4j tends to utilize even with a very small data set, and completely idle: no connections, no queries running. I have Graphaware UUID and APOC plugins configured, otherwise it's pretty out of the box. This on versions 3.2 - 3.4.5. Locally, the java process uses about 1.2-1.6% cpu at all times. On a GCP VM, usage is 2,880 hours, 240 gibibytes in the last rolling month (which is expensive), and that instance goes weeks without a single connection, and then only a short browser session when it is connected.
Note I haven't done any particular performance tuning, allowing memory to auto-config, but we're talking about 99.9% idle (or should be idle) and a 6MB store. Any advice is appreciated!
Greetings,
Putting aside the cpu usage for a moment ... 240GiB of ... what exactly ? Is that network traffic to and from the instance (a 6Mb instance) ? Because either there's constantly big remote queries running (which would contradict what you say) ... or there is some serious data moving to and from that instance (outside of the database). Can you elaborate on that 240GiB number ?
My bad, the confusing GCP billing console does not provide a way to tie a line item to a set of machines. I was mistaking a broken K8s cluster for our Neo4j staging instance.