Hello everyone, I'm back from holidays, i had the time to patch the generator in non-transactionnal mode and it works ! Sadly, the insert rate has been divided by something like... 50 times ? We guessed it would be slower, but didn't thought the gap would be this large.
We will run it once more to have actual numbers/rate for each mode. Anyway, We also thought about writing "Big File" full of insert lines to do some bulk insert but we encountered an unexpected problem. If we insert a new element (be it an edge or a vertex) we can't force manually the @rid. Consequently, we can't write easily all the edges between the vertices because we do not have any way to know in advance the proper vertice's @rids. We did some research, the workaround is to use subqueries in the insert to retrieve the vertices to bind. But that means OrientDB will have to perform two subquery for each edge to insert, it will probably work but we highly doubt it will be faster than the non-transactionnal mode and in the end we would have done this for nothing. Are we, once more, missing something ? Or is writing large insertion file just not a good idea for OrientDB (at least in the case you have lots of edges) ? Thanks a lot and Happy new year ! Cyprien Gottstein. -- --- 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 orient-database+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
