0

I created a script with python to insert 20.000 nodes and 30.000 relationships.

Using the same script, if I import these data in a locally neo4j database created with Neo4j Desktop, I take 1 minutes; on the other side, if I use a neo4j database with Aura, I take 15 minutes.

Why this difference? I have tried with a server with 8GB of memory and 2 CPU.

dav04
  • 19
  • 7
  • That is is quite a big difference. Did you check the Aura console during the import for CPU utilisation? Also, how is the import working? 1 connection, and batching? Otherwise the network in between might play a bigger role.. – bert Apr 09 '21 at 15:52
  • @bert the CPU utilization is always between 20 and 35%, so it's not a lot. I import data using UNWIND but this is not the problem because, before run UNWIND, I have to create the list to pass to it. To create that list, I check every node if already exists inside neo4j, and this step takes a long time with Aura. I have also add indexes but I have the same result. – dav04 Apr 09 '21 at 16:15

0 Answers0