Efficiently getting all out-degrees in a large graph

Neo4j Graph Platform Cypher
1

I have a reasonably large and dense graph of 1.5 million nodes. I'd like to get a list of all out-degrees of relationships of a certain type. I've played around a bit, but the queries I've come up with are all slow enough that I'm looking for feedback on whether I can tune it further.

I'm currently using the following:

MATCH (p:Package)-[r:PACKAGE_DEPENDS_ON]->()
RETURN p.name, COUNT(r) AS count

I'm not actually interested in p.name, but without it, the query just returns a single number. My question is whether there's a way to write this query that doesn't return unnecessary information, and whether that might improve performance?

2
  1. can you share the schema ?
  2. explain plan for the query ?
3

@taobojlen I guess you could use degree centrality by carefully choosing the arguments. In your case, you could specify "OUTGOING" since you only want to compute the out edges. Graph Algorithms in Neo4j: Degree Centrality

4

Yes, here is the schema. I've added the table format since the visualization did not show all labels.

Graph schema

Screenshot from 2020-07-02 09-48-39

  • Red: Version
  • Blue: VersionRequirement
  • Orange: Package
  • Beige: User

Nodes

[
{
  "identity": -3,
  "labels": [
    "VersionRequirement"
  ],
  "properties": {
"indexes": [],
"name": "VersionRequirement",
"constraints": []
  }
}
,
{
  "identity": -2,
  "labels": [
    "User"
  ],
  "properties": {
"indexes": [],
"name": "User",
"constraints": []
  }
}
,
{
  "identity": -4,
  "labels": [
    "Version"
  ],
  "properties": {
"indexes": [],
"name": "Version",
"constraints": []
  }
}
,
{
  "identity": -1,
  "labels": [
    "Package"
  ],
  "properties": {
"indexes": [],
"name": "Package",
"constraints": []
  }
}
]

Relationships

[
{
  "identity": -6,
  "start": -4,
  "end": -4,
  "type": "DEPENDS_ON_RESOLVES_TO",
  "properties": {

  }
}
,
{
  "identity": -8,
  "start": -1,
  "end": -1,
  "type": "PACKAGE_DEPENDS_ON",
  "properties": {

  }
}
,
{
  "identity": -4,
  "start": -3,
  "end": -4,
  "type": "RESOLVES_TO",
  "properties": {

  }
}
,
{
  "identity": -7,
  "start": -4,
  "end": -4,
  "type": "NEXT_VERSION",
  "properties": {

  }
}
,
{
  "identity": -1,
  "start": -4,
  "end": -3,
  "type": "DEPENDS_ON",
  "properties": {

  }
}
,
{
  "identity": -2,
  "start": -2,
  "end": -4,
  "type": "MAINTAINS",
  "properties": {

  }
}
,
{
  "identity": -3,
  "start": -3,
  "end": -1,
  "type": "REQUIREMENT_OF",
  "properties": {

  }
}
,
{
  "identity": -5,
  "start": -4,
  "end": -1,
  "type": "VERSION_OF",
  "properties": {

  }

And the explain plan:
Screenshot from 2020-07-02 09-50-54

5

I might be missing something from this link, but it looks like their query is very similar to the one I have? In my query I do specify a direction for the relationship.

6

Well, running the query today seems totally fine -- yesterday I was waiting for ~30 minutes and assumed that it was due to a badly written query. It seems like Neo4j was stuck in a bad state. Sometimes turning it off and on again does work...!