gregfelice opened a new issue, #2348:
URL: https://github.com/apache/age/issues/2348
## Summary
The Cypher executor's pattern matching (`MATCH (n:Label {key: value})`) does
not leverage GIN or btree indexes on the `properties` agtype column. This
forces sequential scans even when appropriate indexes exist, resulting in
significantly slower property-based lookups compared to native PostgreSQL index
usage.
## Reproduction
Setup: 100K nodes in a graph with a GIN index on the `properties` column.
```sql
-- Create GIN index on properties
CREATE INDEX ON benchmark."Node" USING GIN (properties);
-- This Cypher query does NOT use the index:
SELECT * FROM cypher('benchmark', $$ MATCH (n:Node {bench_id: 42})-[e]->(m)
RETURN m $$) AS (v agtype);
-- EXPLAIN ANALYZE shows Seq Scan on "Node" table
```
## Performance Data
Benchmark W04 (pattern match — find neighbors of a node by property):
| Approach | 10K scale | 100K scale |
|---|---|---|
| Cypher executor (no index use) | ~1ms | ~1ms |
| SQL JOIN with materialized column + btree index | 0.14ms | 0.14ms |
The SDK workaround materializes properties into native PostgreSQL columns
and uses SQL JOINs to bypass the Cypher executor entirely.
## Expected Behavior
`MATCH (n:Label {key: value})` should push property predicates down to
PostgreSQL's index scan when a GIN index exists on the `properties` column,
similar to how `jsonb @> '{"key": "value"}'` leverages GIN indexes.
## Environment
- PostgreSQL 18.2
- Apache AGE 1.7.0 (built from source, branch `release/PG18/1.7.0`)
- 32-core, 64GB RAM, Linux 6.17
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
