Database becomes non-responsive in production

We had issues with long STW pauses on enterprise neo4j running on aws on an i3.8xlarge instance. This occurred even with little pressure on system memory or cpu usage.

We were using OpenJDK. We switched to Oracle JDK and have found far fewer long STW pauses garbage collection. I believe the garbage collection in Oracle's proprietary JDK is superior to the open source version (at least for neo4j). I think we still have some tuning to do of the JVM parameters, but this certainly helped us.

$ java -version
java version "1.8.0_201"
Java(TM) SE Runtime Environment (build 1.8.0_201-b09)
Java HotSpot(TM) 64-Bit Server VM (build 25.201-b09, mixed mode)