PATTERN Cited by 1 source

Parallel multi-slice query

Fetching multiple graph data slices (adjacency lists, property sets) in parallel rather than sequentially, specifically to reduce latency for high-fanout graph traversals where a single hop expands to hundreds or thousands of edges.

Context

In JanusGraph's default getMultiSlices implementation, fetches for multiple slices may be serialized or insufficiently parallelized. For a 4–8 hop identity graph query where each hop fans out to many neighbors, serial slice fetching compounds latency multiplicatively.

Pattern shape

1. Identify multi-slice fetch operations in the graph engine.
2. Execute fetches concurrently (thread pool, async I/O, or batched backend requests).
3. Aggregate results after all parallel fetches complete.
4. Bound parallelism to avoid overwhelming the storage backend.

Seen in

• sources/2026-05-19-airbnb-scaling-identity-graph-unified-knowledge-graph-infrastructure — Airbnb "improved the getMultiSlices interface to fetch data in parallel, significantly reducing latency for high-fanout queries." This was one of three key optimizations to the core JanusGraph engine.

Related

• systems/janusgraph — the engine optimized
• concepts/graph-traversal-fanout — the problem this pattern addresses
• concepts/tail-latency-at-scale — parallel fetching directly reduces tail latency