Graph support

Client includes the executeGraph() method to execute graph queries:

const client = new cassandra.Client({
  contactPoints: ['host1', 'host2'],
  localDataCenter: 'dc1',
  graphOptions: { name: 'demo' }

// executeGraph() method returns a Promise
  .then(function (result) {
    const vertex = result.first();

Alternatively, you can use the callback-based execution:

client.executeGraph('g.V()', function (err, result) {
  const vertex = result.first();
  // ...

Graph Options

You can set default graph options when initializing Client which will be used for all graph statements. For example, to avoid providing a graphName option in each executeGraph() call:

const cassandra = require('cassandra-driver');
const client = new cassandra.Client({
  contactPoints: ['host1', 'host2'],
  localDataCenter: 'dc1',
  graphOptions: { name: 'demo' }

These options may be overridden by specifying the execution profile when calling executeGraph():

// Use a different graph name than the one provided when creating the client instance
const result = await client.executeGraph(query, params, { executionProfile: 'graph-oltp' });
const vertex = result.first();

You can check out more info on Execution Profiles.

Handling Results

Graph queries return a GraphResultSet, which is an iterable of items. The format of the data returned is dependent on the data requested.

Retrieving property values:

const result = await client.executeGraph('g.V().hasLabel("person").values("name")');
for (const name of result) {

Retrieving vertices:

const result = await client.executeGraph('g.V().hasLabel("person")');
for (const vertex of result) {

Retrieving edges:

const result = await client.executeGraph('g.E()');
for (const edge of result) {


Graph traversal execution supports named parameters. Parameters must be passed in as an object:

const traversal = 'g.addV(vertexLabel).property("name", username)';
await client.executeGraph(traversal, { vertexLabel: 'person', username: 'marko' });

Graph types

The DataStax Node.js driver supports a wide variety of TinkerPop types and DSE types. For graph types that don’t have a native JavaScript representation, the driver provides the types module.

For example:

const { types } = require('cassandra-driver');
const { Uuid, InetAddress } = types;

const traversal = 'g.addV("sample").property("uid", uid).property("ip_address", address)';
await client.execute(traversal, { uid: Uuid.random(), address: InetAddress.fromString('') });

The same types are also supported for traversal execution results:

const rs = await client.execute('g.V().hasLabel("sample").values("ip_address")');
for (const ip of rs) {
  console.log(ip instanceof InetAddress); // true

User-defined types

User-defined types (UDTs) are supported in the Node.js driver using JavaScript objects.

const rs = await client.execute('g.V().hasLabel("sample").values("user_address")');
for (const address of rs) {
  console.log(`User address is ${address.street}, ${} ${address.state}`);

In order to use a UDT as a parameter, you must wrap the object instance using asUdt() function to provide additional information to properly represent the UDT on the server.

const { datastax } = require('cassandra-driver');
const { asUdt } = datastax.graph;

// Get the UDT metadata
const udtInfo = await client.metadata.getUdt(graphName, 'address');

// Build the UDT
const address = asUdt({ street: '123 Priam St.', city: 'My City', state: 'MY' }, udtInfo);

const traversal = 'g.addV("sample").property("uid", uid).property("user_address", address)';

// Use the UDT as parameter
await client.execute(traversal, { uid: Uuid.random(), address });