Causal Clustering of Embedded mode database

[nitinjkt]

Hi Team,

We are migrating our inventory system from legacy COTS product to new product that have Neo4j as database.
We are using Neo4j enterprise version 3.5.5
This new product uses neo4j in embedded mode menas our app and database on same machine.
In our deployment architecture we want to deploy application on Kubernetes cluster and
want to keep our database in HA or Causal cluster .
Please let us know how to achieve it .

[david_rosenblum]

Hi @nitinjkt - Neo4j 3.5.5 is pretty old, and upgrading to the latest 3.5.x release at a minimum will make moving to 4.2 easier.
I suggest you look at the following doc to give you ideas on how to transition to using bolt with an embedded app, but you will need to be on 4.2

Neo4j Graph Database Platform

Accessing Neo4j embedded via the Bolt protocol - Neo4j Java Reference

This section describes how to open a Bolt connector to your embedded instance to get GUI administration and other benefits.

If I were you, I would setup minimal app that runs in your K8 and setup a small Neo4j 4.2.x CC and see if things work. Were you not in K8, I would just give you the advice in the manual, that you can transition your app as you see fit.

[nitinjkt]

Hi David,

Thanks for response. We need advise on choosing Embedded or server mode of Neo4j.
While we are talking about micro service architecture application deployed on K8 with huge database size calculation as below

.

What would be best choice of neo4j database mode selection not only in terms of performance but in terms of scalability , manageability.
And is there any roadmap to deprecate embedded mode in future release ?