I have some questions about importing data

(Namestudent) #1

I am a neo4j beginner. I have some questions about importing data.

When we import billions of data, how to use java to import, what better information? What better way? Must they all be converted to CSV format? If converted to csv, is it necessary to pre-organize and modify it according to the specified format?

Thank you

(Kunal Goyal) #2

First you should make sure that what would be your data model according to that you would frame queries for data import.
You should mention that what would be your data source and yes you have to make sure that your system must have the configuration to handle billions of records .

(M. David Allen) #3

There are lots of different ways of importing data into Neo4j.

Probably the most common route is to use CSV, so here's a page giving you full details on that:

Broadly, you can either use the LOAD CSV method or you can use the neo4j-import tool. If you only have a little data, LOAD CSV is easier. If you have lots and lots of data and high speed import is your highest priority, then you want neo4j-import.

(Namestudent) #4

ok!!I'll give it a try.

thank you.

(Benjamin Squire) #5

Fastest way is using neo4j-admin import , this requires formatting all csvs. For billions of nodes its best to use descending ordering if at all possible. See here: https://neo4j.com/docs/operations-manual/current/tools/import/

You can import data in parallel and for approximately 1.8 TB of data I got around 4 hours on a 32 CPU machine with 244 GB ram. You will need a lot of RAM for the import for sure.