We have a testing automation tool that uses graphing as its store. The 
logic behind that is that source code is a tree and trees are graphs. The 
modeling/mapping is pretty straightforward.

Currently we are using neo4j (java api)  and it has two issues:

the more important one is the is the labels one.

The mapping is tight (to preserve space) A value node/vertex can have 
thousands of labels. To harvest those values, we use evaluators (traversers 
for you guys) but GC thrashing happens frequently. 

Do you guys have a ballpark of how many labels any vertex can handle before 
it craps out?

the other one is the elements per write.

The neo4j people suggest around 30,000 per transaction to maximize i/o 
flushing. 

what is your suggestion in this regard?

thanks a lot!

-- 

--- 
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