benchmarks kuzu vs duckdb pgq

[hpvd]

Since there is also a graph extension for duckdb (see https://github.com/cwida/duckpgq-extension)
it would be more than interesting to see some benchmarks kuzu vs duckdb pgq

• to identify strength and weakness of each one
• maybe get pointed to fields to be improved

PS: there are already some comparisons of kuzu vs neo4j
https://thedataquarry.com/blog/embedded-db-2/

[prrao87]

Hi @hpvd are you aware of any examples out there that people are benchmarking this on? The Kuzu vs. Neo4j comparisons were me, but when I wasn't yet part of the Kuzu team :). They're pretty unofficial and informal, so I wonder if there are others out there who are more formally benchmarking it against other graph DBs.

[hpvd]

sorry Im not aware of other benchmarks of Kuzu or other graph DBs vs duckdb pgq.
There are only benchmarks vs neo4J of nearly any graphdb out there.
e.g. falkor (was redis graph) https://benchmark.falkordb.com/

[hpvd]

I think this would be an interesting and important benchmark

• on the one hand because you describe kuzu somehow like duckdb for graph
• and on the other hand because dedicated graph-dbs needs not always to be superior in complex workloads when doing fair comparisons and both systems are well optimised by using indexes see https://www.mdpi.com/2076-3417/12/13/6490

[hpvd]

-> so it could be a great chance for kuzu to shine :-)

[hpvd]

just one more: what about benchmarking kuzu vs falkorDB?
You have pretty different usage profiles but both are claiming highspeeeeed :-)
https://github.com/falkordb/falkordb

It's not only about speed, if others are also good, it's about what could one learn?