Create/Merge getting slower over time

Neo4j Graph Platform Import / Export
1

driver = GraphDatabase.driver(...,
auth=(...))
session = driver.session()
chunk = chunk.drop_duplicates()
for start in range(0, len(chunk), 500):
df_temp = chunk[start:start + 500].copy()
driver.execute_query(item[1], data=df_temp.to_dict('records'))
driver.close()

over time, the operations of create and merge are getting slower. does anyone know the reason? it starts fast but becomes slow over time.

2

@infraplataforma

Any details relative to what version of Neo4j?

also and per MERGE - Cypher Manual

For performance reasons, creating a schema index on the label or property is highly recommended when using MERGE. See Create, show, and delete indexes for more information.

are there indexes to support the merge?

3

im working with auradb and yes, i use constraints and indexes and i tried with merge and create, but it doesn't work well. all options are getting slow over time.

4

@infraplataforma

Aura. And thus Neo4j v5? i presume?

if you preface the MERGE with PROFILE do you see the index being used?

5

correct! and yes, the index is being used

6

@infraplataforma

are you able to share the query plan?

7

Planner COST

Runtime PIPELINED

Runtime version 5.19

Batch size 128

+---------------------+----+--------------------------------------------------------------+----------------+------+---------+----------------+------------------------+-----------+---------------------+
| Operator | Id | Details | Estimated Rows | Rows | DB Hits | Memory (Bytes) | Page Cache Hits/Misses | Time (ms) | Pipeline |
+---------------------+----+--------------------------------------------------------------+----------------+------+---------+----------------+------------------------+-----------+---------------------+
| +ProduceResults | 0 | | 1 | 0 | 0 | 0 | | | |
| | +----+--------------------------------------------------------------+----------------+------+---------+----------------+ | | |
| +EmptyResult | 1 | | 1 | 0 | 0 | | | | |
| | +----+--------------------------------------------------------------+----------------+------+---------+----------------+ | | |
| +Create | 2 | (a)-[r:HAS_PARTNER_RELATION {type: $autoint_2}]->(b) | 1 | 1 | 3 | | | | |
| | +----+--------------------------------------------------------------+----------------+------+---------+----------------+ | | |
| +MultiNodeIndexSeek | 3 | RANGE INDEX a:Company(document) WHERE document = $autoint_0, | 1 | 0 | 0 | 240 | 4/5 | 34.648 | Fused in Pipeline 0 |
| | | RANGE INDEX b:Company(document) WHERE document = $autoint_1 | | | | | | | |
+---------------------+----+--------------------------------------------------------------+----------------+------+---------+----------------+------------------------+-----------+---------------------+

Total database accesses: 3, total allocated memory: 304

im working with millions nodes but the example is with two only