How we can compress data disk footprint in neo4j in bulk import

[manish_iitian4u]

• neo4j version 4.0.2

• neo4j-admin import
  I uploaded same nodes in Neo4j 4.0 and TigerGraph and disk footprint in Neo4j is 115 Gb and Tiger Graph is 30 Gb . So I wanted to know is there way to decrease the disk foot print .

[manish_iitian4u]

Does anyone answer or look at the question posted in community

[roberto1]

you need to check logs, by default neo4j store logs by TX

[clem]

We need more information...

One possibility is that you are storing numbers as Strings instead of Numbers when you do the import. You need to do toFloat() or toInteger() or maybe even toBoolean() before storing the value.

You can check out the types with this call:
RETURN apoc.meta.type(n.supposedNumber)

[clem]

I've been looking at TG. One of their claims is they do compression on everything behind the scenes, whereas Neo4J doesn't. I'm guessing since TG is written in C++, the compression goes very fast. Compression also has the advantage of reducing RAM and Disk footprints, which will also speed things up.

I suspect for Neo4J, you'll have to roll your own data compression for attributes that make the most sense in your use case.

You might have to resort to using APOC compression or tap into various Java library compression