Benchmark results of Neo4j vs Dgraph

[Lazhar Ichir]

Just thought you guys would be interested into this benchmark of Neo4j versus Dgraph:

• Dgraph is 160x faster than Neo4j for loading graph data.
• Dgraph consumes 5x lesser memory compared to Neo4j and is at least 3x faster for intertwined reads and writes.
• For intertwined reads and writes, Dgraph is at least 3x to 6x faster.

What do you guys think?

L

[Michael Hunger]

The benchmark has several big flaws, and shouldn't be taken seriously at least for the Neo4j side.

Otherwise what Jim said on your GH issue:

Jim Webber: I think vendor benchmarks aren't worth the paper they're written on. This one doesn't even distinguish between caching results (neo4j doesn't) and query plans (it certainly does). Their setup of neo4j is (deliberately?) naive too. On the whole, my position is that they need to prove themselves in the marketplace and not on their blog.

As you can probably imagine, we don't have the capacity to help other vendors with their benchmarks or teaching them about Neo4j basics.

Usually, we recommend that people interested in Neo4j take their own data and model & import into the Neo4j property graph and run their own use-cases against them. We're glad to help real projects to be successful and happy with their Neo4j use, which seems to work very well so far.

If you're interested in learning more about Neo4j feel free to join our London user group and come to one of our frequent events.

Cheers, Michael

--
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+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[ArchenROOT]

Hi guys, although older post, still relevant.

I mean dgraph benchmark is open source, so anyone can go and customize the benchmark neo4j configuration, so it performs best, but noone did it :-) 

"we don't have the capacity to help other vendors with their benchmarks or teaching them about Neo4j basics" - this is especially what you should do when one is showing with open source data his solution does better. I wouldn't call it vendor benchmark as soon as it is fully open source and they welcome feedback.

I think soon we will need something like http://www.tpc.org/tpch/ for graph databases, some standardized set of different type of graphs and common graph algorithms to be executed.

To unsubscribe from this group and stop receiving emails from it, send an email to neo4j+un...@googlegroups.com.