Tuning Neo4j for Performance

Question

I have imported data using Michael Hunger's Batch Import, through which I created:-

4,612,893 nodes
14,495,063 properties
    node properties are indexed.
5,300,237 relationships

{Question} Cypher queries are executing too slow almost crawling, simple traversal is taking > 5 mins to return resultset, Please let me know how to tune the server to get better performance and what I am doing wrong.

Store Details:-

-rw-r--r-- 1 root root 567M Jul 12 12:42 data/graph.db/neostore.propertystore.db
-rw-r--r-- 1 root root 167M Jul 12 12:42 data/graph.db/neostore.relationshipstore.db
-rw-r--r-- 1 root root  40M Jul 12 12:42 data/graph.db/neostore.nodestore.db
-rw-r--r-- 1 root root 7.8M Jul 12 12:42 data/graph.db/neostore.propertystore.db.strings
-rw-r--r-- 1 root root  330 Jul 12 12:42 data/graph.db/neostore.propertystore.db.index.keys
-rw-r--r-- 1 root root  292 Jul 12 12:42 data/graph.db/neostore.relationshiptypestore.db.names
-rw-r--r-- 1 root root  153 Jul 12 12:42 data/graph.db/neostore.propertystore.db.arrays
-rw-r--r-- 1 root root   88 Jul 12 12:42 data/graph.db/neostore.propertystore.db.index
-rw-r--r-- 1 root root   69 Jul 12 12:42 data/graph.db/neostore
-rw-r--r-- 1 root root   58 Jul 12 12:42 data/graph.db/neostore.relationshiptypestore.db
-rw-r--r-- 1 root root    9 Jul 12 12:42 data/graph.db/neostore.id
-rw-r--r-- 1 root root    9 Jul 12 12:42 data/graph.db/neostore.nodestore.db.id
-rw-r--r-- 1 root root    9 Jul 12 12:42 data/graph.db/neostore.propertystore.db.arrays.id
-rw-r--r-- 1 root root    9 Jul 12 12:42 data/graph.db/neostore.propertystore.db.id
-rw-r--r-- 1 root root    9 Jul 12 12:42 data/graph.db/neostore.propertystore.db.index.id
-rw-r--r-- 1 root root    9 Jul 12 12:42 data/graph.db/neostore.propertystore.db.index.keys.id
-rw-r--r-- 1 root root    9 Jul 12 12:42 data/graph.db/neostore.propertystore.db.strings.id
-rw-r--r-- 1 root root    9 Jul 12 12:42 data/graph.db/neostore.relationshipstore.db.id
-rw-r--r-- 1 root root    9 Jul 12 12:42 data/graph.db/neostore.relationshiptypestore.db.id
-rw-r--r-- 1 root root    9 Jul 12 12:42 data/graph.db/neostore.relationshiptypestore.db.names.id

I am using

neo4j-community-1.9.1
java version "1.7.0_25"
Amazon EC2 m1.large instance with Ubuntu 12.04.2 LTS (GNU/Linux 3.2.0-40-virtual x86_64)
RAM ~8GB.
EBS 200 GB, neo4j is running on EBS volume.

Invoked as ./neo4j-community-1.9.1/bin/neo4j start

Below are the neo4j server info:

neostore.nodestore.db.mapped_memory                 161M
neostore.relationshipstore.db.mapped_memory         714M
neostore.propertystore.db.mapped_memory              90M
neostore.propertystore.db.index.keys.mapped_memory    1M
neostore.propertystore.db.strings.mapped_memory     130M
neostore.propertystore.db.arrays.mapped_memory      130M
mapped_memory_page_size                               1M
all_stores_total_mapped_memory_size                 500M

---

{Data Model} is like Social Graph :-

User-User
    User-[:FOLLOWS]->User
User-Item
    User-[:CREATED]->Item
    User-[:LIKE]->Item
    User-[:COMMENT]->Item
    User-[:VIEW]->Item
Cluster-User
    User-[:FACEBOOK]->SocialLogin_Cluster
Cluster-Item
    Item-[:KIND_OF]->Type_Cluster
Cluster-Cluster
    Cluster-[:KIND_OF]->Type_Cluster

{Some Queries} and time:

START u=node(467242)
MATCH u-[r1:LIKE|COMMENT]->a<-[r2:LIKE|COMMENT]-lu-[r3:LIKE]-b
WHERE NOT(a=b)
RETURN u,COUNT(b)

Query took 1015348ms. Returned 70956115 result count.

START a=node:nodes(kind="user")
RETURN a,length(a-[:CREATED|LIKE|COMMENT|FOLLOWS]-()) AS cnt
ORDER BY cnt DESC
LIMIT 10

Query took 231613ms

---

From the suggestions, I upgraged the box to M1.xlarge and M2.2xlarge

• M1.xlarge (vCPU:4,ECU:8,RAM:15 GB,Instance Storage:~600 GB)
• M2.2xlarge (vCPU:4,ECU:13,RAM:34 GB,Instance Storage:~800 GB)

I tuned the properties like below, and running from instance storage (as against EBS)

neo4j.properties

neostore.nodestore.db.mapped_memory=1800M
neostore.relationshipstore.db.mapped_memory=1800M
neostore.propertystore.db.mapped_memory=100M
neostore.propertystore.db.strings.mapped_memory=150M
neostore.propertystore.db.arrays.mapped_memory=10M

neo4j-wrapper.conf

wrapper.java.additional.1=-d64
wrapper.java.additional.1=-server
wrapper.java.additional=-XX:+UseConcMarkSweepGC
wrapper.java.additional=-XX:+CMSClassUnloadingEnabled
wrapper.java.initmemory=4098
wrapper.java.maxmemory=8192

but still the queries (like below) run in minutes ~5-8 minutes, which is not acceptable from recommendation point of view.

Query:

START u=node(467242)
MATCH u-[r1:LIKE]->a<-[r2:LIKE]-lu-[r3:LIKE]-b
RETURN u,COUNT(b)

---

{Profiling}

neo4j-sh (0)$ profile START u=node(467242) MATCH u-[r1:LIKE|COMMENT]->a<-[r2:LIKE|COMMENT]-lu-[r3:LIKE]-b RETURN u,COUNT(b);
==> +-------------------------+
==> | u            | COUNT(b) |
==> +-------------------------+
==> | Node[467242] | 70960482 |
==> +-------------------------+
==> 1 row
==> 
==> ColumnFilter(symKeys=["u", "  INTERNAL_AGGREGATEad2ab10d-cfc3-48c2-bea9-be4b9c1b5595"], returnItemNames=["u", "COUNT(b)"], _rows=1, _db_hits=0)
==> EagerAggregation(keys=["u"], aggregates=["(  INTERNAL_AGGREGATEad2ab10d-cfc3-48c2-bea9-be4b9c1b5595,Count)"], _rows=1, _db_hits=0)
==>   TraversalMatcher(trail="(u)-[r1:LIKE|COMMENT WHERE true AND true]->(a)<-[r2:LIKE|COMMENT WHERE true AND true]-(lu)-[r3:LIKE WHERE true AND true]-(b)", _rows=70960482, _db_hits=71452891)
==>     ParameterPipe(_rows=1, _db_hits=0)


neo4j-sh (0)$ profile START u=node(467242) MATCH u-[r1:LIKE|COMMENT]->a<-[r2:LIKE|COMMENT]-lu-[r3:LIKE]-b RETURN count(distinct a),COUNT(distinct b),COUNT(*);
==> +--------------------------------------------------+
==> | count(distinct a) | COUNT(distinct b) | COUNT(*) |
==> +--------------------------------------------------+
==> | 1950              | 91294             | 70960482 |
==> +--------------------------------------------------+
==> 1 row
==> 
==> ColumnFilter(symKeys=["  INTERNAL_AGGREGATEe6b94644-0a55-43d9-8337-491ac0b29c8c", "  INTERNAL_AGGREGATE1cfcd797-7585-4240-84ef-eff41a59af33", "  INTERNAL_AGGREGATEea9176b2-1991-443c-bdd4-c63f4854d005"], returnItemNames=["count(distinct a)", "COUNT(distinct b)", "COUNT(*)"], _rows=1, _db_hits=0)
==> EagerAggregation(keys=[], aggregates=["(  INTERNAL_AGGREGATEe6b94644-0a55-43d9-8337-491ac0b29c8c,Distinct)", "(  INTERNAL_AGGREGATE1cfcd797-7585-4240-84ef-eff41a59af33,Distinct)", "(  INTERNAL_AGGREGATEea9176b2-1991-443c-bdd4-c63f4854d005,CountStar)"], _rows=1, _db_hits=0)
==>   TraversalMatcher(trail="(u)-[r1:LIKE|COMMENT WHERE true AND true]->(a)<-[r2:LIKE|COMMENT WHERE true AND true]-(lu)-[r3:LIKE WHERE true AND true]-(b)", _rows=70960482, _db_hits=71452891)
==>     ParameterPipe(_rows=1, _db_hits=0)

Please let me know the configuration and neo4j startup arguments for tuning. Thanks in advance

Answer 1

Running this on my macbook air with little RAM and CPU with your dataset.

You will get much faster than my results with more memory mapping, GCR cache and more heap for caches. Also make sure to use parameters in your queries.

You are running into combinatorial explosion.

Every step of the path adds "times rels" elements/rows to your matched subgraphs.

See for instance here: you end up at 269268 matches but you only have 81674 distinct lu's

The problem is that for each row the next match is expanded. So if you use distinct in between to limit the sizes again it will be some order of magnitutes less data. Same for the next level.

START u=node(467242)
MATCH u-[:LIKED|COMMENTED]->a
WITH distinct a
MATCH a<-[r2:LIKED|COMMENTED]-lu
RETURN count(*),count(distinct a),count(distinct lu);

+---------------------------------------------------+
| count(*) | count(distinct a) | count(distinct lu) |
+---------------------------------------------------+
| 269268   | 1952              | 81674              |
+---------------------------------------------------+
1 row

895 ms

START u=node(467242)
MATCH u-[:LIKED|COMMENTED]->a
WITH distinct a
MATCH a<-[:LIKED|COMMENTED]-lu
WITH distinct lu
MATCH lu-[:LIKED]-b
RETURN count(*),count(distinct lu), count(distinct b)
;
+---------------------------------------------------+
| count(*) | count(distinct lu) | count(distinct b) |
+---------------------------------------------------+
| 2311694  | 62705              | 91294             |
+---------------------------------------------------+

Here you have 2.3M total matches and only 91k distinct elements. So almost 2 orders of magnitude.

This is a huge aggregation which is rather a BI / statistics query that an OLTP query. Usually you can store the results e.g. on the user-node and only re-execute this in the background.

THESE kind of queries are again global graph queries (statistics / BI ), in this case top 10 users.

Usually you would run these in the background (e.g. once per day or hour) and connect the top 10 user nodes to a special node or index that then can be queried in a few ms.

START a=node:nodes(kind="user") RETURN count(*);
+----------+
| count(*) |
+----------+
| 3889031  |
+----------+
1 row

27329 ms

After all you are running a match across the whole graph, i.e. 4M users that's a graph global, not a graph local query.

START n=node:nodes(kind="top-user")
MATCH n-[r?:TOP_USER]-()
DELETE r
WITH distinct n
START a=node:nodes(kind="user")
MATCH a-[:CREATED|LIKED|COMMENTED|FOLLOWS]-()
WITH n, a,count(*) as cnt
ORDER BY cnt DESC
LIMIT 10
CREATE a-[:TOP_USER {count:cnt} ]->n;

+-------------------+
| No data returned. |
+-------------------+
Relationships created: 10
Properties set: 10
Relationships deleted: 10

70316 ms

The querying would then be:

START n=node:nodes(kind="top-user")
MATCH n-[r:TOP_USER]-a
RETURN a, r.count
ORDER BY r.count DESC;

+--------------------------------------------------------------------------------------------------------------------------------------------------------------+
| a                                                                                                                                                  | r.count |
+--------------------------------------------------------------------------------------------------------------------------------------------------------------+
….
+--------------------------------------------------------------------------------------------------------------------------------------------------------------+
10 rows

4 ms

Answer 2

Okay, so first of all, for only 8GB of memory that's a very large graph. You should seriously consider getting a larger box. Neo4j actually provides an extremely nice hardware calculator that will let you determine exactly what is appropriate for your needs:

http://neotechnology.com/calculatorv2/

In a contrived way (since there are more relevant metrics to determining size) their calculator estimates should should be dedicating about 10GB at a minimum.

Secondly, Neo4j and any graph database will have issues with nodes that have a large number of connections. If you're looking to tune your instance to perform better (after getting a bigger box) I would suggest looking for any massive nodes with a large number of connections as those will seriously impact performance.

After seeing your examples I'm quite certain you've got a graph with a number of nodes that have a much larger number of connections that other nodes. This will inherently slow down your performance. You might also try more narrow queries. Especially when you're already working on a server that's too small you don't want to run the kind of extremely taxing large return queries you've got there.

There are some things about your queries that could be cleaned up, but I really urge you to get the appropriately sized box for your graph and actually do some introspection into the number of connections your most connected nodes have.

It also looks like you have an artificial cap on your Java Heap size. If you try starting up java with a command like:

java -Xmx8g //Other stuff

You'll allocate 8 gigs instead of the standard ~500 Megs, which would also help.