Index

Overview

Indexing, or property indexing, is a technique used in Ultipa to accelerate the retrieval of nodes and edges with specific properties. By avoiding full graph scans, indexes enable the database to quickly locate relevant data. This is especially advantageous when working with large graphs.

Index Types

Ultipa supports single index on one property and composite index which involve multiple properties from a schema.

Showing Indexes

To retrieve indexes in the current graphset:

// Shows all indexes
show().index() 

// Shows all node indexes
show().node_index()

// Shows all edge indexes
show().edge_index()

The information about indexes is organized into the _nodeIndex or _edgeIndex table. Each table provides essential details about each index:

Creating Indexes

You can create one or more indexes using a single create() statement. Each index is created by chaining a node_index() or edge_index() method. Note that each property can only have one single index. The index creation runs as a job, you may run show().job(<id?>) afterward to verify the success of the creation.

System properties in Ultipa are inherently optimized for query performance and have built-in efficiencies. They do not support indexing.

create()
  .node_index(@<schema>.<property>(<bytes?>), "<indexName>")
  .edge_index(@<schema>.<property>(<bytes?>), "<indexName>")
  .node_index(@<schema>(<property1>(<bytes1?>), <property2>(<bytes2?>), ...), "<indexName>")
  .edge_index(@<schema>(<property1>(<bytes1?>), <property2>(<bytes2?>), ...), "<indexName>")
  ...

^[1] In standard English text, most encodings (such as ASCII or UTF-8) use 1 byte per character. However, for non-English characters, the byte size may vary—for example, one Chinese character typically occupies 3 bytes.

To create single index named cBalance for the property balance of card nodes:

create().node_index(@card.balance, "cBalance")

To create single index named name for the property name (string type) of card nodes, restricting the indexed byte-length as 10:

create().node_index(@card.name(10), "name")

To create composite index named transAmountNotes for properties amount and notes (text type, restricting the indexed byte-lengt as 10) for transfer edges:

create().edge_index(@transfer(amount, notes(10)), "transAmountNotes")

To create multiple indexes:

create()
  .node_index(@card.balance, "balance")
  .edge_index(@transfer(amount, notes(10)), "transAmountNotes")

Dropping Indexes

You can drop one or more indexes using a single drop() statement. Each index is specified by chaining a node_index() or edge_index() method. Dropping an index does not affect the actual property values stored in shards.

A property with an index cannot be dropped until the index is deleted.

To drop the node index cBalance:

drop().node_index("cBalance")

To drop the edge index transAmountNotes:

drop().edge_index("transAmountNotes")

To drop multiple indexes:

drop().node_index("balance").edge_index("transAmountNotes")

Using Indexes

Applicable Queries

Indexes are automatically applied when the corresponding properties are used in the following types of queries. They are not effective in other types of queries.

1. Node retrieval using find().nodes(). For example,

create().node_index(@user.age, "user_age_index")

The user_age_index is effective in the following queries:

find().nodes({@user.age == 45}) as n return n

find().nodes({age > 45}) as n return n

In the second query, the node schema is not specified, so user_age_index is only partially used during the search for user nodes.

2. Edge retrieval using find().edges(). For example,

create().edge_index(@links.weight, "links_weight_index")

The links_weight_index is effective in the following query:

find().edges({@links.weight == 2}) as e return e

3. Start node filtering in path patterns.

The above user_age_index is effective in the following query:

n({@user.age > 45}).e().n().e().n() as p return p

It does not apply to the following query:

n().e().n().e().n({@user.age > 45}) as p return p

Leftmost Prefix Rule

The order of properties in a composite index matters — queries that match the leftmost properties of the index (i.e., the first property or the first few properties in the defined order) will benefit from the index.

For example:

create().node_index(@user(name(10),age), 'name_age')

• find().nodes({@user.name == "Kavi" && @user.age > 20}) uses the index.
• find().nodes({@user.name == "Kavi"}) uses the index.
• find().nodes({@user.age > 20}) doesn't use the index.
• find().nodes({@user.name == "Kavi" && @user.age > 20 && @user.grade == 7}) uses the index, meanwhile it contains the filtering for the @user.grade property which lacks an index.

String Byte-Length Limitation

When using indexes with string or text properties, ensure the byte-length of the string used in the filter does not exceed the defined limit when creating the index.

For example, an index Username is created for the name property of the user nodes with a 8-byte limitation:

create().node_index(@user.name(8), "Username")

The query below won't utilize the Username index as the specified string Aventurine exceeds the 8-byte limit:

find().nodes({@user.name == "Aventurine"}) as n return n