Why does execution of profile change when variable length path increases

[meshanen]

Hi here is my code

MATCH (parent:PART {category:'SKU',part_id:'400695-01'})
MATCH (child:PART)-[r:belongs_to*0..10]->(parent)
return parent.part_id, count(child.part_id) as part_count

This gives me the execution profile on the right, which takes very long

when I lower the varlength path to 9, the query suddenly becomes much faster

MATCH (parent:PART {category:'SKU',part_id:'400695-01'})
MATCH (child:PART)-[r:belongs_to*0..9]->(parent)
return parent.part_id, count(child.part_id) as part_count
and now the execution profile becomes the one on the left which is much faster.

Anyone knows why the execution plan changes this drastically when I change the variable length path from 10 to 9.

[tard_gabriel (Gabriel Tardif)]

A possible answer would be that the number of children at the deep 10 is much more than 9 depending on the way your data are structured.