INSERT & BULK INSERT

Insert single rows with INSERT INTO or multiple rows efficiently with BULK INSERT INTO. For tables with embedding columns, vectors are computed automatically.

INSERT Syntax

SQL

INSERT INTO table_name (col1, col2, ...) VALUES (val1, val2, ...);

Returns the number of rows inserted (1 for a single INSERT).

SQL

INSERT INTO users (id, name, email) VALUES (1, 'Alice', 'alice@example.com');
INSERT INTO users (id, name, email) VALUES (2, 'Bob', 'bob@example.com');

BULK INSERT

BULK INSERT is an AuroraSQL extension for inserting multiple rows in a single statement. It is significantly faster than multiple individual INSERTs because it batches WAL writes and embedding requests.

SQL

BULK INSERT INTO table_name (col1, col2, ...) VALUES
  (val1, val2, ...),
  (val3, val4, ...),
  (val5, val6, ...);

SQL

BULK INSERT INTO users (id, name, email, age) VALUES
  (3, 'Charlie', 'charlie@example.com', 35),
  (4, 'Diana', 'diana@example.com', 28),
  (5, 'Eve', 'eve@example.com', 42);

Tip

Use BULK INSERT when inserting more than ~10 rows. For 1M rows, BULK INSERT achieves 20,267 rows/sec in embedded mode vs ~454 rows/sec for individual INSERTs over the wire protocol.

Automatic Embeddings

For tables with EMBEDDING MODEL columns, embeddings are computed automatically on INSERT. You provide the text value; GalaxDB handles the rest.

SQL

CREATE TABLE docs (
    id   INT PRIMARY KEY,
    body TEXT EMBEDDING MODEL 'sentence-transformers/all-MiniLM-L6-v2' DIM 384
);

-- Just insert the text — embedding computed automatically
INSERT INTO docs (id, body) VALUES (1, 'machine learning and neural networks');
INSERT INTO docs (id, body) VALUES (2, 'cooking recipes italian pasta');

-- BULK INSERT also computes embeddings automatically
BULK INSERT INTO docs (id, body) VALUES
  (3, 'rust programming language systems'),
  (4, 'deep learning transformers attention'),
  (5, 'database storage engine LSM tree');

The embedding computation is asynchronous — the INSERT returns immediately after the row is written to the WAL. The sidecar processes the embedding queue in the background. For SEMANTIC_MATCH queries, rows with pending embeddings are excluded until their embeddings are computed.

Examples

Insert with all column types

SQL

CREATE TABLE products (
    id       INT PRIMARY KEY,
    name     TEXT NOT NULL,
    price    FLOAT,
    in_stock BOOL,
    sku      TEXT
);

INSERT INTO products (id, name, price, in_stock, sku)
VALUES (1, 'Widget Pro', 29.99, true, 'WGT-001');

Python client INSERT

Python

import galaxdb

db = galaxdb.Database("./data")

# Single insert — returns row count (int)
count = db.execute("INSERT INTO docs (id, body) VALUES (1, 'hello world')")
print(count)  # 1

# Bulk insert
count = db.execute("""
    BULK INSERT INTO docs (id, body) VALUES
      (2, 'machine learning'),
      (3, 'deep learning'),
      (4, 'natural language processing')
""")
print(count)  # 3

psycopg2 parameterized INSERT

Python

import psycopg2

conn = psycopg2.connect(host="localhost", port=5433, dbname="galaxdb", sslmode="disable")
cur = conn.cursor()

# Parameterized insert (safe against SQL injection)
cur.execute(
    "INSERT INTO docs (id, body) VALUES (%s, %s)",
    (1, "machine learning and neural networks")
)
conn.commit()

CREATE TABLE

SELECT