Suitability of XTDB for graph queries

[tarek-sater-pf]

Hi, we are evaluating XTDB for constructing a temporal property graph.

The docs mention that it is somewhat graph-like:

XTDB is not the first database to offer this level of support for dynamic data, but it is the first to do so with well-defined columnar data types via Apache Arrow. The flexibility of this approach is ideal for underpinning highly-normalized, graph-like data models.

but no mention whether it is suitable for recursive queries or graph traversal. Is this possible in XTDB, and if so what are the limitations and performance implications?

Most queries are going to be less than 10 hops. In addition, these queries may be temporal, over a big time range.

[refset]

Hey @tarek-sater-pf - sorry I missed this question on here, but thanks for re-posting on Discord (clearly I need to create a Discord bot so I only have one place to check for notifications on these things 🙈 )

Pasting my response from there, for posterity:

great question 🙂 the graph angle of XTDB hasn't had much attention from us in recent months, admittedly, but last time someone asked about this was here: https://discuss.xtdb.com/t/xtdbs-sweet-spot/52/9?u=refset - you can see there some examples for DFS/BFS with SQL CTEs
XTDB v1 was more naturally pointed towards solving graph problems because of the WCOJ engine, the (inferred) 6NF data model and the expressive nature of the Datalog, but in v2 the situation is quite different. We do have ideas in mind to accelerate graph workloads in the v2 engine, but nothing has been built specifically to tackle those challenges, so YMMV.