Optimize Your Workload on Neo4j Aura

This information is appropriate for Aura Professional and Aura Enterprise customers. If you follow this guidance, you will utilize your AuraDB resources efficiently.

First, you should understand how best to optimize your queries.  There is a great article on Performance Tuning With Neo4j AuraDB.  You can request Aura Support to provide you with long running queries to focus your tuning activities.  Please do take this action.  Having sub-optimal queries can be very inefficient.

Second, you have the benefit of the (1 leader, 2 followers) architecture in Aura Professional and Aura Enterprise to distribute workload.  This does not happen automatically: you need to take action to mark your read queries.  Please follow the information in this article, to Explicitly Mark Read Queries. In some cases, we see 80% of the work on the leader even though the writes/reads split 40/60.  Again, you need to take action to distribute the workload.

Third, please follow best practices for max connections.  You should review this article, How to operate with many applications connecting to my AuraDB Instance using the driver (maximum connections).  Using too many connections can overwhelm an instance, and create an imbalance in CPU usage.