Starburst Platform

Starburst Galaxy is a managed Trino service providing serverless distributed SQL query execution. BenchBox provides first-class Starburst support, inheriting Trino’s SQL dialect and federated query capabilities.

Features

• Managed Trino - No cluster management required
• Trino dialect - Inherits SQL dialect from Trino
• Federated queries - Query data across multiple sources
• Built-in catalogs - Pre-configured data connectors
• HTTPS secure - Always encrypted connections
• Multiple table formats - Iceberg, Hive, Delta Lake support

Quick Start

The commands install the Trino driver, set credentials, and run a benchmark.

uv add trino

export STARBURST_HOST=my-cluster.trino.galaxy.starburst.io
export STARBURST_USER=joe@example.com/accountadmin
export STARBURST_PASSWORD=your-password

benchbox run --platform starburst --benchmark tpch --scale 1.0

Authentication

Starburst Galaxy uses a unique username format that combines your email with a role:

username = email/role
# Example: joe@example.com/accountadmin

Configuration Methods

Environment Variables (recommended): the first three variables are the credentials. Optionally, configure the role separately: set STARBURST_USER without the role suffix and set STARBURST_ROLE, which is appended automatically. In that case the second STARBURST_USER line replaces the first. STARBURST_CATALOG optionally sets the default catalog.

export STARBURST_HOST=my-cluster.trino.galaxy.starburst.io
export STARBURST_USER=joe@example.com/accountadmin
export STARBURST_PASSWORD=your-password

export STARBURST_USER=joe@example.com
export STARBURST_ROLE=accountadmin

export STARBURST_CATALOG=tpch_sf1

benchbox run --platform starburst --benchmark tpch --scale 1.0

Credentials file: the CLI also reads these settings from the starburst section of ~/.benchbox/credentials.yaml. Values written as ${VAR} are read from the environment. Environment variables apply only to settings the file leaves unset.

starburst:
  host: my-cluster.trino.galaxy.starburst.io
  username: joe@example.com/accountadmin
  password: ${STARBURST_PASSWORD}
  catalog: tpch_sf1

Starburst has no --platform-option settings for connection details, catalog, schema, or table format.

Configuration Options

Set each option in the credentials file, or through its environment variable where one exists.

Option	Environment Variable	Required	Default	Description
host	STARBURST_HOST	Yes	-	Galaxy cluster hostname
username	STARBURST_USER / STARBURST_USERNAME	Yes	-	User email or email/role
password	STARBURST_PASSWORD	Yes	-	Password or API key
role	STARBURST_ROLE	No	-	Role (appended to username if not included)
catalog	STARBURST_CATALOG	No	-	Default catalog
port	STARBURST_PORT	No	443	HTTPS port
schema	-	No	from benchmark and scale	Schema for benchmark tables
table_format	-	No	memory	Table format: memory, hive, iceberg, delta
verify_ssl	-	No	true	SSL certificate verification

Usage Examples

Basic Benchmark

The commands run TPC-H at scale factor 1 and TPC-DS at scale factor 10.

benchbox run --platform starburst --benchmark tpch --scale 1.0

benchbox run --platform starburst --benchmark tpcds --scale 10.0

With Specific Catalog

export STARBURST_CATALOG=my_catalog

benchbox run --platform starburst --benchmark tpch --scale 1.0

To set the schema as well, add schema: benchmark_data to the starburst section of the credentials file.

With Table Format

Tables use the memory format unless the credentials file sets table_format. For Iceberg tables:

starburst:
  table_format: iceberg

Use delta for Delta Lake tables.

Python API

from benchbox import TPCH
from benchbox.platforms.starburst import StarburstAdapter

adapter = StarburstAdapter(
    host="my-cluster.trino.galaxy.starburst.io",
    username="joe@example.com/accountadmin",
    password="your-password",
    catalog="tpch_catalog",
    schema="benchmark_data",
    table_format="iceberg",
)

benchmark = TPCH(scale_factor=1.0)
benchmark.generate_data()
adapter.load_benchmark(benchmark)
results = adapter.run_benchmark(benchmark)

Architecture

Starburst inherits from Trino, which means:

• SQL Dialect: Uses Trino’s SQL dialect for query translation
• Connector Syntax: Same catalog.schema.table naming convention
• Session Properties: Trino session properties are supported

get_platform_family returns "trino", and get_inherited_platform also returns "trino".

from benchbox.core.platform_registry import PlatformRegistry

family = PlatformRegistry.get_platform_family("starburst")

parent = PlatformRegistry.get_inherited_platform("starburst")

Table Formats

Starburst Galaxy supports multiple table formats:

Format	Description	Use Case
memory	In-memory tables	Fast testing, small data
hive	Hive format	Compatibility with Hive ecosystem
iceberg	Apache Iceberg	Production analytics, ACID transactions
delta	Delta Lake	Databricks ecosystem integration

Iceberg is recommended for analytics. Set table_format: iceberg in the starburst section of the credentials file.

Comparison: Starburst vs Trino

Feature	Starburst Galaxy	Self-Hosted Trino
Deployment	Cloud managed	Self-hosted cluster
Authentication	Password/API key	Configurable
SSL	Always HTTPS	Configurable
Scaling	Automatic	Manual
Catalogs	Pre-configured	Manual setup
Cost	Pay-per-use	Infrastructure cost
Best For	Quick start, production	Full control, customization

When to Use Starburst

Use Starburst when:

• You need managed Trino without cluster management
• Running federated queries across multiple data sources
• Using Iceberg/Delta Lake table formats
• You want built-in catalog management

Use self-hosted Trino instead when:

• You need full control over cluster configuration
• Cost optimization is critical
• You have existing Trino infrastructure
• You need specific Trino plugins/connectors

Troubleshooting

Authentication Failed (401)

Starburst Galaxy authentication failed.

Solutions:

1. Verify username format: email/role (e.g., joe@example.com/accountadmin)
2. Check password is correct
3. Verify credentials at galaxy.starburst.io

Connection Refused

Cannot connect to Starburst Galaxy at {host}:{port}

Solutions:

1. Verify host is correct: {cluster-name}.trino.galaxy.starburst.io
2. Check network connectivity
3. Verify no firewall blocking port 443

SSL Certificate Error

SSL certificate error connecting to Starburst Galaxy.

Solutions:

1. Check network proxy settings
2. Verify SSL certificate chain
3. Set verify_ssl: false in the starburst section of ~/.benchbox/credentials.yaml (not recommended for production)

Catalog Not Found

Catalog 'my_catalog' does not exist

Solutions:

1. Create the catalog in Starburst Galaxy console
2. Use an existing catalog: check available catalogs in the console
3. Leave the catalog unset to use the default

Related Documentation

• Trino Platform - Self-hosted Trino benchmarking
• Deployment Modes Guide - Platform deployment architecture
• Platform Selection Guide - Choose the right platform
• Getting Started - Quick start guide