How were you able to automate the execution of the neo4j-admin commands? I was googling around and searching this forum for the answer to my question. Based on the quote above, I think you may have the answer to my question.
You also state that,
When adding collections of edges/nodes which would add a significant percentage of new nodes/edges to a database it can be faster to rebuild from scratch than to add them. In my experience it has always been faster. The bulk loader is lightning fast in comparison to load csv.
Now that my database has grown to include 500,000,000 relationships, I am finding this to be completely true and accurate. I've detailed my struggle here: