3 ms·
Yes a graph database will happily lead you down a n^3 (or worse!) path when trying to query for a single relation if you are not wise about your indexes, etc.
by justonceokay 7mo ago
Yes a graph database will happily lead you down a n^3 (or worse!) path when trying to query for a single relation if you are not wise about your indexes, etc.
- adsharma 7mo agoAre you talking about the query plan for scanning the rel table? Kuzu used a hash index and a join. Trying to make it optional. Try explain match (a)-[b]->(c) return a.rowid, b.rowid, c.rowid;
- cluckindan 7mo agoThat sounds like a ”graph” DB which implements edges as separate tables, like building a graph in a standard SQL RDB. If you wish to avoid that particular caveat, look for a graph DB which materializes edges within vertices/nodes. The obvious caveat there is that the edges are not normalized, which may or may not be an issue for your particulat application.