> > On Wednesday, February 19, 2014 6:12:19 PM UTC-8, David Erickson wrote: >> >> Thanks for the response Max, unfortunately our expectation is that 1600 >> vertices/s of writing capability is a few orders of magnitude below what we >> expect/need. Probably your benchmark also has a lot of overhead from >> making web calls and Ruby itself. What are the expectations of what Neo4j >> ought to be able to handle if calling it natively (embedded) from >> multithreaded Java, with some optimal number of batched new Vertices/edges >> per TX commit? >> >> Just for some higher numbers... This is over the transactional Cypher HTTP endpoint using my go driver, from a single threaded client. CREATE statements at ~25k/sec w/ 10k or 1k tx size, sent in batches of 100. MERGE on unique constraint statements at ~2.5k/sec, best tx size tested was 100, sent all in one batch.
Benchmarks from today on my macbook pro, using 2.0.1: https://gist.github.com/wfreeman/ab808dc5244da441fc58 Code here: https://github.com/wfreeman/cq/blob/master/driver_bench_test.go The best I've gotten without BatchInserter or batch graphdb, using embedded Java, is shy of 50k nodes/sec (again, single threaded client). I am curious--what kind of numbers are you guys looking for? Wes -- You received this message because you are subscribed to the Google Groups "Neo4j" group. To unsubscribe from this group and stop receiving emails from it, send an email to neo4j+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.