cypher - Neo4j - query performance degradation -

here query:

match (p:publisher)-[r:published]->(w:woka)-[s:authored]-(a:author)  match (l:language)-[t:used]->(w:woka)  (a.author_name =~ '.*camus.*' , a.author_name =~ '.*albert.*')  return p.publisher_name, w.woka_title, a.author_name, l.language_name; 

the first time executing result returned in 3.8 seconds. second execution couple minutes later result returned in 15.1 seconds. more executing longer response time. third execution response time increasing , several moments later getting results between 30 , 90 seconds.

i user of (development) database. no data added or deleted or changed there. no indexes dropped or created there. when closing 2 out of 3 connections database response time goes 15 seconds.

memory set 4gb init , max 8gb. server has 16gb total memory.

what happening here? why response time differs much?

how big graph? allocates lot of heap caching , there not enough space running queries without garbage collection.

i presume relationships 1:n, if not add with distinct p,w,a in between 2 matches.

your query suboptimal, , create lot of intermediate results, can use profile @ query plan.

try this:

profile match (a:author) (a.author_name =~ '.*camus.*' , a.author_name =~ '.*albert.*')  match (p:publisher)-[:published]->(w:woka)<-[:authored]-(a) match (l:language)-[:used]->(w:woka)  return p.publisher_name, w.woka_title, a.author_name, l.language_name;