Insert, update, or delete Spanner Graph data

Note: This feature is available with the Spanner Enterprise edition and Enterprise Plus edition. For more information, see the Spanner editions overview.

This document describes how to mutate data in Spanner Graph using the Google Cloud console, the Google Cloud CLI, and client libraries. Mutating graph data includes inserting, updating, and deleting nodes and edges in Spanner Graph.

Spanner Graph maps data from tables to graph nodes and edges. To mutate data in a graph, you must mutate data in the corresponding input tables. For more information, see the Spanner Graph schema overview.

Before you begin

To complete the Google Cloud console steps and the code samples in this document, you must first follow the steps in Set up and query Spanner Graph to do the following:

1. Create an instance.
2. Create a database with a Spanner schema.
3. Insert graph data.

Insert nodes or edges

To insert nodes or edges, use the Google Cloud console, the Google Cloud CLI, or the Spanner client libraries to insert rows into node or edge tables.

In the Google Cloud console and in the Google Cloud CLI, you can use GoogleSQL Data Manipulation Language (DML) to insert. In the Spanner client library, you can use DML or Mutation APIs.

Before you insert an edge, make sure that the source and destination nodes connected by the edge exist. If you insert an edge when the source or destination node connected by the edge doesn't exist, you might get referential integrity violation errors. For more information, see Missing source node violates INTERLEAVE IN relationship and Missing destination node violates foreign key constraint.

The following examples insert Account nodes and Transfer edges into the graph:

Console

1. Run statements in the Google Cloud console.

2. In the Google Cloud console, enter the following DML statement and click Run Query:

-- Insert 2 Account nodes.
INSERT INTO Account (id, create_time, is_blocked)
VALUES (1, CAST('2000-08-10 08:18:48.463959-07:52' AS TIMESTAMP), false);
INSERT INTO Account (id, create_time, is_blocked)
VALUES (2, CAST('2000-08-12 07:13:16.463959-03:41' AS TIMESTAMP), true);

-- Insert 2 Transfer edges.
INSERT INTO AccountTransferAccount (id, to_id, create_time, amount)
VALUES (1, 2, CAST('2000-09-11 03:11:18.463959-06:36' AS TIMESTAMP), 100);
INSERT INTO AccountTransferAccount (id, to_id, create_time, amount)
VALUES (1, 1, CAST('2000-09-12 04:09:34.463959-05:12' AS TIMESTAMP), 200);

gcloud

1. Execute statements with the gcloud CLI.
2. In the gcloud CLI, run the following commands:

gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="INSERT INTO Account (id, create_time, is_blocked) VALUES (1, CAST('2000-08-10 08:18:48.463959-07:52' AS TIMESTAMP), false)"
gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="INSERT INTO Account (id, create_time, is_blocked) VALUES (2, CAST('2000-08-12 07:13:16.463959-03:41'  AS TIMESTAMP), true)"
gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="INSERT INTO AccountTransferAccount (id, to_id, create_time, amount) VALUES (1, 2, CAST('2000-09-11 03:11:18.463959-06:36' AS TIMESTAMP), 100)"
gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="INSERT INTO AccountTransferAccount (id, to_id, create_time, amount) VALUES (1, 1, CAST('2000-09-12 04:09:34.463959-05:12' AS TIMESTAMP), 200)"

Client libraries

Python

def insert_data_with_dml(instance_id, database_id):
    """Inserts sample data into the given database using a DML statement."""

    spanner_client = spanner.Client()
    instance = spanner_client.instance(instance_id)
    database = instance.database(database_id)

    def insert_accounts(transaction):
        row_ct = transaction.execute_update(
            "INSERT INTO Account (id, create_time, is_blocked) "
            "  VALUES"
            "    (1, CAST('2000-08-10 08:18:48.463959-07:52' AS TIMESTAMP), false),"
            "    (2, CAST('2000-08-12 07:13:16.463959-03:41' AS TIMESTAMP), true)"
        )

        print("{} record(s) inserted into Account.".format(row_ct))

    def insert_transfers(transaction):
        row_ct = transaction.execute_update(
            "INSERT INTO AccountTransferAccount (id, to_id, create_time, amount) "
            "  VALUES"
            "    (1, 2, CAST('2000-09-11 03:11:18.463959-06:36' AS TIMESTAMP), 100),"
            "    (1, 1, CAST('2000-09-12 04:09:34.463959-05:12' AS TIMESTAMP), 200) "
        )

        print("{} record(s) inserted into AccountTransferAccount.".format(row_ct))

    database.run_in_transaction(insert_accounts)
    database.run_in_transaction(insert_transfers)

Java

static void insertUsingDml(DatabaseClient dbClient) {
  dbClient
      .readWriteTransaction()
      .run(
          transaction -> {
            String sql =
                "INSERT INTO Account (id, create_time, is_blocked) "
                    + "  VALUES"
                    + "    (1, CAST('2000-08-10 08:18:48.463959-07:52' AS TIMESTAMP), false),"
                    + "    (2, CAST('2000-08-12 07:13:16.463959-03:41' AS TIMESTAMP), true)";
            long rowCount = transaction.executeUpdate(Statement.of(sql));
            System.out.printf("%d record(s) inserted into Account.\n", rowCount);
            return null;
          });

  dbClient
      .readWriteTransaction()
      .run(
          transaction -> {
            String sql =
                "INSERT INTO AccountTransferAccount (id, to_id, create_time, amount) "
                    + "  VALUES"
                    + "    (1, 2, CAST('2000-09-11 03:11:18.463959-06:36' AS TIMESTAMP), 100),"
                    + "    (1, 1, CAST('2000-09-12 04:09:34.463959-05:12' AS TIMESTAMP), 200) ";
            long rowCount = transaction.executeUpdate(Statement.of(sql));
            System.out.printf("%d record(s) inserted into AccountTransferAccount.\n", rowCount);
            return null;
          });
}

Go

import (
	"context"
	"fmt"
	"io"

	"cloud.google.com/go/spanner"
)

func insertGraphDataWithDml(w io.Writer, db string) error {
	ctx := context.Background()
	client, err := spanner.NewClient(ctx, db)
	if err != nil {
		return err
	}
	defer client.Close()

	// Execute a ReadWriteTransaction to insert values into the 'Account' table
	// underpinning 'Account' nodes in 'FinGraph'. The function run by ReadWriteTransaction
	// executes an 'INSERT' SQL DML statement. Graph queries run after this
	// transaction is committed will observe the effects of the new 'Account's
	// added to the graph.
	_, err1 := client.ReadWriteTransaction(ctx, func(ctx context.Context, txn *spanner.ReadWriteTransaction) error {
		stmt := spanner.Statement{
			SQL: `INSERT INTO Account (id, create_time, is_blocked)
            		VALUES
            	    	(1, CAST('2000-08-10 08:18:48.463959-07:52' AS TIMESTAMP), false),
            			(2, CAST('2000-08-12 07:13:16.463959-03:41' AS TIMESTAMP), true)`,
		}
		rowCount, err := txn.Update(ctx, stmt)
		if err != nil {
			return err
		}
		fmt.Fprintf(w, "%d Account record(s) inserted.\n", rowCount)
		return err
	})

	if err1 != nil {
		return err1
	}

	// Execute a ReadWriteTransaction to insert values into the 'AccountTransferAccount'
	// table underpinning 'AccountTransferAccount' edges in 'FinGraph'. The function run
	// by ReadWriteTransaction executes an 'INSERT' SQL DML statement.
	// Graph queries run after this transaction is committed will observe the effects
	// of the edges added to the graph.
	_, err2 := client.ReadWriteTransaction(ctx, func(ctx context.Context, txn *spanner.ReadWriteTransaction) error {
		stmt := spanner.Statement{
			SQL: `INSERT INTO AccountTransferAccount (id, to_id, create_time, amount)
					VALUES
						(1, 2, CAST('2000-09-11 03:11:18.463959-06:36' AS TIMESTAMP), 100),
						(1, 1, CAST('2000-09-12 04:09:34.463959-05:12' AS TIMESTAMP), 200)`,
		}
		rowCount, err := txn.Update(ctx, stmt)
		if err != nil {
			return err
		}
		fmt.Fprintf(w, "%d AccountTransferAccount record(s) inserted.\n", rowCount)
		return err
	})

	return err2
}

C++

void InsertDataWithDml(google::cloud::spanner::Client client) {
  using ::google::cloud::StatusOr;
  namespace spanner = ::google::cloud::spanner;

  std::int64_t rows_inserted;
  auto commit_result = client.Commit(
      [&client, &rows_inserted](
          spanner::Transaction txn) -> StatusOr<spanner::Mutations> {
        auto insert =
            client.ExecuteDml(std::move(txn), spanner::SqlStatement(R"""(
          INSERT INTO Account (id, create_time, is_blocked)
          VALUES
          (1, CAST('2000-08-10 08:18:48.463959-07:52' AS TIMESTAMP), false),
          (2, CAST('2000-08-12 07:13:16.463959-03:41' AS TIMESTAMP), true)
        )"""));
        if (!insert) return std::move(insert).status();
        rows_inserted = insert->RowsModified();
        return spanner::Mutations{};
      });
  if (!commit_result) throw std::move(commit_result).status();
  std::cout << "Rows inserted into Account: " << rows_inserted << "\n";

  commit_result = client.Commit(
      [&client, &rows_inserted](
          spanner::Transaction txn) -> StatusOr<spanner::Mutations> {
        auto insert =
            client.ExecuteDml(std::move(txn), spanner::SqlStatement(R"""(
          INSERT INTO AccountTransferAccount (id, to_id, create_time, amount)
          VALUES
          (1, 2, CAST('2000-09-11 03:11:18.463959-06:36' AS TIMESTAMP), 100),
          (1, 1, CAST('2000-09-12 04:09:34.463959-05:12' AS TIMESTAMP), 200)
        )"""));
        if (!insert) return std::move(insert).status();
        rows_inserted = insert->RowsModified();
        return spanner::Mutations{};
      });
  if (!commit_result) throw std::move(commit_result).status();
  std::cout << "Rows inserted into AccountTransferAccount: " << rows_inserted
            << "\n";

  std::cout << "Insert was successful [spanner_insert_graph_data_with_dml]\n";
}

Update nodes or edges

To update existing nodes or edges, use the Google Cloud console, the gcloud CLI, or the Spanner client libraries.

You can update existing nodes or edges using a GoogleSQL Data Manipulation Language (DML) statement, or Spanner Graph queries with a DML statement. In the Spanner client library, you can also use Mutation APIs.

Update nodes or edges with DML

The following examples update an Account node and a Transfer edge in the graph using DML:

Console

1. Run statements in the Google Cloud console.

2. In the Google Cloud console, enter the following DML statement and click Run Query:

-- Update Account node
UPDATE Account SET is_blocked = false WHERE id = 2;

-- Update Transfer edge
UPDATE AccountTransferAccount
SET amount = 300
WHERE id = 1 AND to_id = 2;

Console

1. Execute statements with the gcloud CLI.
2. In the gcloud CLI, run the following commands:

gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="UPDATE Account SET is_blocked = false WHERE id = 2"
gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="UPDATE AccountTransferAccount SET amount = 300 WHERE id = 1 AND to_id = 2"

Client libraries

Python

def update_data_with_dml(instance_id, database_id):
    """Updates sample data from the database using a DML statement."""

    spanner_client = spanner.Client()
    instance = spanner_client.instance(instance_id)
    database = instance.database(database_id)

    def update_accounts(transaction):
        row_ct = transaction.execute_update(
            "UPDATE Account SET is_blocked = false WHERE id = 2"
        )

        print("{} Account record(s) updated.".format(row_ct))

    def update_transfers(transaction):
        row_ct = transaction.execute_update(
            "UPDATE AccountTransferAccount SET amount = 300 WHERE id = 1 AND to_id = 2"
        )

        print("{} AccountTransferAccount record(s) updated.".format(row_ct))

    database.run_in_transaction(update_accounts)
    database.run_in_transaction(update_transfers)

Java

static void updateUsingDml(DatabaseClient dbClient) {
  dbClient
      .readWriteTransaction()
      .run(
          transaction -> {
            String sql = "UPDATE Account SET is_blocked = false WHERE id = 2";
            long rowCount = transaction.executeUpdate(Statement.of(sql));
            System.out.printf("%d Account record(s) updated.\n", rowCount);
            return null;
          });

  dbClient
      .readWriteTransaction()
      .run(
          transaction -> {
            String sql =
                "UPDATE AccountTransferAccount SET amount = 300 WHERE id = 1 AND to_id = 2";
            long rowCount = transaction.executeUpdate(Statement.of(sql));
            System.out.printf("%d AccountTransferAccount record(s) updated.\n", rowCount);
            return null;
          });
}

Go

import (
	"context"
	"fmt"
	"io"

	"cloud.google.com/go/spanner"
)

func updateGraphDataWithDml(w io.Writer, db string) error {
	ctx := context.Background()
	client, err := spanner.NewClient(ctx, db)
	if err != nil {
		return err
	}
	defer client.Close()

	// Execute a ReadWriteTransaction to update the 'Account' table underpinning
	// 'Account' nodes in 'FinGraph'. The function run by ReadWriteTransaction
	// executes an 'UPDATE' SQL DML statement. Graph queries run after this
	// transaction is committed will observe the effects of the update to 'Account'
	// with 'id' = 2.
	_, err1 := client.ReadWriteTransaction(ctx, func(ctx context.Context, txn *spanner.ReadWriteTransaction) error {
		stmt := spanner.Statement{
			SQL: `UPDATE Account SET is_blocked = false WHERE id = 2`,
		}
		rowCount, err := txn.Update(ctx, stmt)
		if err != nil {
			return err
		}
		fmt.Fprintf(w, "%d Account record(s) updated.\n", rowCount)
		return err
	})

	if err1 != nil {
		return err1
	}

	// Execute a ReadWriteTransaction to update the 'AccountTransferAccount' table
	// underpinning 'AccountTransferAccount' edges in 'FinGraph'. The function run
	// by ReadWriteTransaction executes an 'UPDATE' SQL DML statement.
	// Graph queries run after this transaction is committed will observe the effects
	// of the update to 'AccountTransferAccount' where the source of the transfer has
	// 'id' 1 and the destination has 'id' 2.
	_, err2 := client.ReadWriteTransaction(ctx, func(ctx context.Context, txn *spanner.ReadWriteTransaction) error {
		stmt := spanner.Statement{
			SQL: `UPDATE AccountTransferAccount SET amount = 300 WHERE id = 1 AND to_id = 2`,
		}
		rowCount, err := txn.Update(ctx, stmt)
		if err != nil {
			return err
		}
		fmt.Fprintf(w, "%d AccountTransferAccount record(s) updated.\n", rowCount)
		return err
	})

	return err2
}

C++

void UpdateDataWithDml(google::cloud::spanner::Client client) {
  using ::google::cloud::StatusOr;
  namespace spanner = ::google::cloud::spanner;

  auto commit_result = client.Commit(
      [&client](spanner::Transaction txn) -> StatusOr<spanner::Mutations> {
        auto update = client.ExecuteDml(
            std::move(txn),
            spanner::SqlStatement(
                "UPDATE Account SET is_blocked = false WHERE id = 2"));
        if (!update) return std::move(update).status();
        return spanner::Mutations{};
      });
  if (!commit_result) throw std::move(commit_result).status();

  commit_result = client.Commit(
      [&client](spanner::Transaction txn) -> StatusOr<spanner::Mutations> {
        auto update =
            client.ExecuteDml(std::move(txn), spanner::SqlStatement(R"""(
          UPDATE AccountTransferAccount
            SET amount = 300 WHERE id = 1 AND to_id = 2)"""));
        if (!update) return std::move(update).status();
        return spanner::Mutations{};
      });
  if (!commit_result) throw std::move(commit_result).status();

  std::cout << "Update was successful [spanner_update_graph_data_with_dml]\n";
}

Update nodes or edges with graph queries and DML

The following examples update an Account node and a Transfer edge in the graph using Spanner Graph queries with DML:

Console

1. Run statements in the Google Cloud console.

2. In the Google Cloud console, enter the following Spanner Graph query with a DML statement, then click Run Query:

-- Use Graph pattern matching to identify Account nodes to update:
UPDATE Account SET is_blocked = false
WHERE id IN {
  GRAPH FinGraph
  MATCH (a:Account WHERE a.id = 1)-[:TRANSFERS]->{1,2}(b:Account)
  RETURN b.id
}

gcloud

1. Execute statements with the gcloud CLI.
2. In the gcloud CLI, run the following commands:

gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="UPDATE Account SET is_blocked = false"
gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="UPDATE AccountTransferAccount SET amount = 300 WHERE id = 1 AND to_id = 2"
    --sql=" WHERE id IN { GRAPH FinGraph MATCH (a:Account WHERE a.id = 1)-[:TRANSFERS]->{1,2}(b:Account) RETURN b.id }"

Client libraries

Python

def update_data_with_graph_query_in_dml(instance_id, database_id):
    """Updates sample data from the database using a DML statement."""

    spanner_client = spanner.Client()
    instance = spanner_client.instance(instance_id)
    database = instance.database(database_id)

    def update_accounts(transaction):
        row_ct = transaction.execute_update(
            "UPDATE Account SET is_blocked = true "
            "WHERE id IN {"
            "  GRAPH FinGraph"
            "  MATCH (a:Account WHERE a.id = 1)-[:TRANSFERS]->{1,2}(b:Account)"
            "  RETURN b.id}"
        )

        print("{} Account record(s) updated.".format(row_ct))

    database.run_in_transaction(update_accounts)

Java

static void updateUsingGraphQueryInDml(DatabaseClient dbClient) {
  dbClient
      .readWriteTransaction()
      .run(
          transaction -> {
            String sql =
                "UPDATE Account SET is_blocked = true "
                    + "WHERE id IN {"
                    + "  GRAPH FinGraph"
                    + "  MATCH (a:Account WHERE a.id = 1)-[:TRANSFERS]->{1,2}(b:Account)"
                    + "  RETURN b.id}";
            long rowCount = transaction.executeUpdate(Statement.of(sql));
            System.out.printf("%d Account record(s) updated.\n", rowCount);
            return null;
          });
}

Go

import (
	"context"
	"fmt"
	"io"

	"cloud.google.com/go/spanner"
)

func updateGraphDataWithGraphQueryInDml(w io.Writer, db string) error {
	ctx := context.Background()
	client, err := spanner.NewClient(ctx, db)
	if err != nil {
		return err
	}
	defer client.Close()

	// Execute a ReadWriteTransaction to update the 'Account' table underpinning
	// 'Account' nodes in 'FinGraph'. The function run by ReadWriteTransaction
	// executes an 'UPDATE' SQL DML statement. Graph queries run after this
	// transaction is committed will observe the effects of the updates to 'Account's
	//
	// The update is performed for all 'Account's whose 'id' is returned by
	// the graph query in the 'IN' subquery, i.e., all 'Account's that have
	// received transfers directly or via one intermediary from an 'Account'
	// whose 'id' is 1.
	_, err = client.ReadWriteTransaction(ctx, func(ctx context.Context, txn *spanner.ReadWriteTransaction) error {
		stmt := spanner.Statement{
			SQL: `UPDATE Account SET is_blocked = true 
            	  WHERE id IN {
            	    GRAPH FinGraph 
            	    MATCH (a:Account WHERE a.id = 1)-[:TRANSFERS]->{1,2}(b:Account)
            	    RETURN b.id}`,
		}
		rowCount, err := txn.Update(ctx, stmt)
		if err != nil {
			return err
		}
		fmt.Fprintf(w, "%d Account record(s) updated.\n", rowCount)
		return err
	})

	return err
}

C++

void UpdateDataWithGraphQueryInDml(google::cloud::spanner::Client client) {
  using ::google::cloud::StatusOr;
  namespace spanner = ::google::cloud::spanner;
  auto commit_result = client.Commit(
      [&client](spanner::Transaction txn) -> StatusOr<spanner::Mutations> {
        auto update =
            client.ExecuteDml(std::move(txn), spanner::SqlStatement(R"""(
              UPDATE Account SET is_blocked = true
              WHERE id IN {
                GRAPH FinGraph
                MATCH (a:Account WHERE a.id = 1)-[:TRANSFERS]->{1,2}(b:Account)
                RETURN b.id})"""));
        if (!update) return std::move(update).status();
        return spanner::Mutations{};
      });
  if (!commit_result) throw std::move(commit_result).status();
  std::cout << "Update was successful "
            << "[spanner_update_graph_data_with_graph_query_in_dml]\n";
}

Delete nodes or edges

To delete existing nodes or edges, use the Google Cloud console, the gcloud CLI, or the Spanner client libraries.

In the Google Cloud console and the gcloud CLI, you can use GoogleSQL Data Manipulation Language (DML) to delete. In the Spanner client library, you can use DML or Mutation APIs.

Before you delete a node, make sure no edges exist that refer to the node. If these types of edges exist, you might get referential integrity violation errors. For more information, see Orphaned outgoing edge violates parent-child relationship and Orphaned incoming edge violates parent-child relationship.

The following examples delete a Transfer edge and an Account node from the graph.

Console

1. Run statements in the Google Cloud console.
2. In the Google Cloud console, enter the following DML statement and click Run Query:

-- Delete Transfer edge
DELETE FROM AccountTransferAccount
WHERE id = 1 AND to_id = 2;

-- Delete Account node
DELETE FROM Account WHERE id = 2;

gcloud

1. Execute statements with the gcloud CLI.
2. In the gcloud CLI, run the following commands:

gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="DELETE FROM AccountTransferAccount WHERE id = 1 AND to_id = 2"
gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="DELETE FROM Account WHERE id = 2"

Client libraries

Python

def delete_data_with_dml(instance_id, database_id):
    """Deletes sample data from the database using a DML statement."""

    spanner_client = spanner.Client()
    instance = spanner_client.instance(instance_id)
    database = instance.database(database_id)

    def delete_transfers(transaction):
        row_ct = transaction.execute_update(
            "DELETE FROM AccountTransferAccount WHERE id = 1 AND to_id = 2"
        )

        print("{} AccountTransferAccount record(s) deleted.".format(row_ct))

    def delete_accounts(transaction):
        row_ct = transaction.execute_update("DELETE FROM Account WHERE id = 2")

        print("{} Account record(s) deleted.".format(row_ct))

    database.run_in_transaction(delete_transfers)
    database.run_in_transaction(delete_accounts)

Java

static void deleteUsingDml(DatabaseClient dbClient) {
  dbClient
      .readWriteTransaction()
      .run(
          transaction -> {
            String sql = "DELETE FROM AccountTransferAccount WHERE id = 1 AND to_id = 2";
            long rowCount = transaction.executeUpdate(Statement.of(sql));
            System.out.printf("%d AccountTransferAccount record(s) deleted.\n", rowCount);
            return null;
          });

  dbClient
      .readWriteTransaction()
      .run(
          transaction -> {
            String sql = "DELETE FROM Account WHERE id = 2";
            long rowCount = transaction.executeUpdate(Statement.of(sql));
            System.out.printf("%d Account record(s) deleted.\n", rowCount);
            return null;
          });
}

Go

// Copyright 2024 Google LLC
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//     https://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

package spanner


import (
	"context"
	"fmt"
	"io"

	"cloud.google.com/go/spanner"
)

func deleteGraphDataWithDml(w io.Writer, db string) error {
	ctx := context.Background()
	client, err := spanner.NewClient(ctx, db)
	if err != nil {
		return err
	}
	defer client.Close()

	// Execute a ReadWriteTransaction to update the 'AccountTransferAccount'
	// table underpinning 'AccountTransferAccount' edges in 'FinGraph'. The
	// function run by ReadWriteTransaction executes an 'DELETE' SQL DML
	// statement. This has the effect of deleting the 'AccountTransferAccount'
	// edge where the source 'id' is 1 and the destination 'id' is 2 from the graph.
	_, err1 := client.ReadWriteTransaction(ctx, func(ctx context.Context, txn *spanner.ReadWriteTransaction) error {
		stmt := spanner.Statement{SQL: `DELETE FROM AccountTransferAccount WHERE id = 1 AND to_id = 2`}
		rowCount, err := txn.Update(ctx, stmt)
		if err != nil {
			return err
		}
		fmt.Fprintf(w, "%d AccountTransferAccount record(s) deleted.\n", rowCount)
		return nil
	})

	if err1 != nil {
		return err1
	}

	// Execute a ReadWriteTransaction to update the 'Account' table underpinning
	//'Account' nodes in 'FinGraph'. In 'FinGraph', nodes can only be deleted
	// after any edges referencing the nodes have been deleted first. The function
	// run by ReadWriteTransaction executes an 'DELETE' SQL DML statement. This has
	// the effect of deleting the 'Account' node whose 'id' is 1 from the graph.
	_, err2 := client.ReadWriteTransaction(ctx, func(ctx context.Context, txn *spanner.ReadWriteTransaction) error {
		stmt := spanner.Statement{SQL: `DELETE FROM Account WHERE id = 2`}
		rowCount, err := txn.Update(ctx, stmt)
		if err != nil {
			return err
		}
		fmt.Fprintf(w, "%d Account record(s) deleted.\n", rowCount)
		return nil
	})

	return err2
}

C++

void DeleteDataWithDml(google::cloud::spanner::Client client) {
  using ::google::cloud::StatusOr;
  namespace spanner = ::google::cloud::spanner;

  auto commit_result = client.Commit([&client](spanner::Transaction txn)
                                         -> StatusOr<spanner::Mutations> {
    auto deleted = client.ExecuteDml(
        std::move(txn),
        spanner::SqlStatement(
            "DELETE FROM AccountTransferAccount WHERE id = 1 AND to_id = 2"));
    if (!deleted) return std::move(deleted).status();
    return spanner::Mutations{};
  });
  if (!commit_result) throw std::move(commit_result).status();

  commit_result = client.Commit(
      [&client](spanner::Transaction txn) -> StatusOr<spanner::Mutations> {
        auto deleted = client.ExecuteDml(
            std::move(txn),
            spanner::SqlStatement("DELETE FROM Account WHERE id = 2"));
        if (!deleted) return std::move(deleted).status();
        return spanner::Mutations{};
      });
  if (!commit_result) throw std::move(commit_result).status();

  std::cout << "Delete was successful [spanner_delete_graph_data_with_dml]\n";
}

You can combine Spanner Graph queries with your DML statement, as shown in the following example:

  -- Use Graph pattern matching to identify Account nodes to delete:
  DELETE FROM AccountTransferAccount
  WHERE id IN {
    GRAPH FinGraph
    MATCH (a:Account WHERE a.id = 1)-[:TRANSFERS]->(b:Account)
    RETURN b.id
  }

Options for mutating graph data

You can automatically mutate graph data in the following ways:

• Automatically delete edges in the graph by using the ON DELETE CASCADE action.
• Automatically delete nodes and edges in the graph using the TTL policy. For more information, see TTL on nodes and edges.

To efficiently bulk update and delete nodes and edges in the graph, use partitioned DML.

What's next

• Read the Spanner Graph queries overview.
• Learn best practices for tuning Spanner Graph queries.