Seeking advice from anyone experienced with importing data that does not 
fit the ETL tool use case (as we understand it).  Our data is based on 
conventional graph data structures:

1.) A set of node.csv files each representing unique properties for each 
class type
2.) A set of edge.csv files with heterogeneous sources and targets, e.g. 
nodes are not of the same class type (mixed).

To be clear, these are not datasets coming from RDBMS structures (PK/FK 
constructs).  Our usual sequence for import is step 1 - load all nodes 
(each having a defined unique id, step 2, create edges among nodes based on 
defined node ids.

It appears from google searching others have faced a similar issue and 
there are no apparent solutions.  We have tried, without success, the 
following:

ETL tool - not suitable (the csv example in documentation is not compatible 
with the above model); it appears you can not configure an "edge only" load 
file(?).
Gremlin & BatchGraph - works for Titan, Neo4j without issue; does not work 
with OrientDB since you can not assign node ids to be referenced for the 
getVertex call.

This has been a surprising development since it took us less than an hour 
to do the same in Neo4J.  Three days later with OrientDB, not a single edge 
record has been created.

We are hopeful this is a case of typical "end user" ignorance.  Aside from 
previous postings here, in Stackoverflow, or otherwise found with extensive 
google searching, has anyone successfully established a method for 
importing data with this use case?

Thanks in advance for any insights!

P.S. We do not have the capability to handle the requirement in Java, hence 
a scripted solution is our only option (SQL, Gremlin, etc.)

-- 

--- 
You received this message because you are subscribed to the Google Groups 
"OrientDB" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to