StarRocks Platform

StarRocks is an open-source MPP (Massively Parallel Processing) columnar OLAP database, hosted under the Linux Foundation. It is designed for sub-second analytics on large-scale datasets. BenchBox connects via the MySQL protocol using PyMySQL (port 9030) and supports Stream Load via the BE HTTP API (port 8040) for high-throughput data ingestion. SQLGlot provides official starrocks dialect support for SQL translation.

StarRocks is used in production by Airbnb, Alibaba, Coinbase, Pinterest, and Tencent, among others. It delivers competitive performance against ClickHouse, Trino, and Druid on analytical workloads.

Features

MySQL protocol connectivity - Standard MySQL wire protocol via PyMySQL (port 9030)
Stream Load HTTP API - High-throughput data ingestion via BE HTTP port 8040
Full TPC-H support - All 22 queries with row count validation
Full TPC-DS support - All 99 queries with row count validation
Official SQLGlot dialect - Native starrocks dialect for accurate SQL translation
Columnar storage - Duplicate Key model optimized for OLAP workloads
Distributed hash partitioning - Automatic data distribution across nodes
Tuning support - Partitioning, sorting, and distribution tuning at table creation time

Quick Start

Install the PyMySQL dependency directly or through the StarRocks extra. StarRocks must be running before you configure the connection and run TPC-H:

uv add pymysql

uv add benchbox --extra starrocks

export STARROCKS_HOST=localhost
export STARROCKS_PORT=9030

benchbox run --platform starrocks --benchmark tpch --scale 0.01

Docker Quick Start

The Docker image runs both the FE and BE. After it starts, verify connectivity with the mysql client, then run the benchmark:

docker run -p 9030:9030 -p 8030:8030 -p 8040:8040 \
    starrocks/allin1-ubuntu:latest

mysql -h 127.0.0.1 -P 9030 -u root -e "SELECT 1"

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

Configuration Options

Option	Environment Variable	Default	Description
`host`	`STARROCKS_HOST`	`localhost`	StarRocks FE hostname
`port`	`STARROCKS_PORT`	`9030`	MySQL protocol port (FE)
`username`	`STARROCKS_USER`	`root`	Database username
`password`	`STARROCKS_PASSWORD`	`""`	Database password
`database`	`STARROCKS_DATABASE`	auto-generated	Target database name
`http_port`	`STARROCKS_HTTP_PORT`	`8040`	BE HTTP port for Stream Load
`max_execution_time`	-	`300`	Query timeout in seconds
`disable_result_cache`	-	`true`	Disable query cache for accurate benchmarking
`strict_validation`	-	`true`	Enable strict data validation
`deployment_mode`	-	`self-hosted`	Deployment mode identifier

The first six options are also --platform-option names. max_execution_time, disable_result_cache, strict_validation, and deployment_mode are not; pass them to StarRocksAdapter in Python.

Data Loading

BenchBox loads data into StarRocks using batch INSERT statements via the MySQL protocol. The adapter handles both TPC pipe-delimited (.tbl) and standard CSV formats automatically.

Loading Process

Schema creation - Tables are created with StarRocks-optimized DDL (Duplicate Key model, hash distribution)
Type conversion - DuckDB/standard SQL types are mapped to StarRocks equivalents (e.g., HUGEINT to LARGEINT, TEXT to VARCHAR(65533))
Batch inserts - Data is loaded in batches of 1,000 rows using parameterized INSERT INTO ... VALUES statements
Constraint handling - Foreign keys are removed (StarRocks does not enforce them); primary keys are converted to Duplicate Key model

Stream Load (HTTP API)

StarRocks provides a high-throughput Stream Load API on the BE HTTP port (default 8040). While the current adapter uses batch INSERT for broad compatibility, the Stream Load endpoint is available for custom ingestion workflows:

Endpoint: http://<be_host>:8040/api/<database>/<table>/_stream_load
Protocol: HTTP PUT with CSV or JSON payload
Throughput: Significantly higher than row-by-row INSERT for large datasets

Type Mappings

Source Type	StarRocks Type
`HUGEINT`	`LARGEINT`
`INTEGER`	`INT`
`STRING` / `TEXT`	`VARCHAR(65533)`
`DOUBLE PRECISION`	`DOUBLE`
`REAL`	`FLOAT`
`BLOB`	`VARCHAR(65533)`
`DECIMAL(p,s)`	`DECIMAL(p,s)` (unchanged)
`BOOLEAN`	`BOOLEAN` (unchanged)

Usage Examples

Basic Benchmarks

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

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

benchbox run --platform starrocks --benchmark tpch --queries Q1,Q6,Q17

Environment Variable Configuration

export STARROCKS_HOST=starrocks-fe.example.com
export STARROCKS_PORT=9030
export STARROCKS_USER=benchbox
export STARROCKS_PASSWORD=secret
export STARROCKS_DATABASE=benchmark_db
export STARROCKS_HTTP_PORT=8040

benchbox run --platform starrocks --benchmark tpch --scale 10.0

Custom Database Name

benchbox run --platform starrocks --benchmark tpch --scale 1.0 \
    --platform-option database=my_benchmarks

Dry Run (Preview)

benchbox run --platform starrocks --benchmark tpch --scale 1.0 --dry-run ./preview

Architecture

Adapter Structure

The StarRocks adapter is composed of four mixins plus the main adapter class:

Module	Class	Responsibility
`adapter.py`	`StarRocksAdapter`	Main adapter entry point, data integrity validation
`setup.py`	`StarRocksSetupMixin`	Connection management, database lifecycle (create/drop)
`workload.py`	`StarRocksWorkloadMixin`	Schema creation, data loading, query execution
`tuning.py`	`StarRocksTuningMixin`	Benchmark configuration, session settings, table tuning
`metadata.py`	`StarRocksMetadataMixin`	Platform info, CLI arguments, dialect selection

Connection Model

BenchBox CLI
    |
    v
StarRocksAdapter
    |
    +-- MySQL Protocol (PyMySQL) --> FE (port 9030)
    |       - Schema DDL
    |       - Query execution
    |       - Batch INSERT loading
    |
    +-- HTTP API (Stream Load) --> BE (port 8040)
            - High-throughput data ingestion

Platform Information

At runtime, BenchBox captures platform metadata:

{
    "platform_type": "starrocks",
    "platform_name": "StarRocks",
    "deployment_mode": "self-hosted",
    "configuration": {
        "host": "localhost",
        "port": 9030,
        "database": "benchbox_tpch_sf1",
        "http_port": 8040
    },
    "platform_version": "3.x.x",
    "client_library_version": "1.x.x"
}

Tuning and Optimization

Automatic Benchmark Configuration

The adapter automatically applies session-level optimizations when running benchmarks:

Query timeout: Set via query_timeout (default: 300 seconds)
Query cache: Disabled by default (enable_query_cache = false) for accurate timing
Planner timeout: Extended for complex OLAP queries (new_planner_optimize_timeout = 30000)
Profiling: Disabled during benchmarks (enable_profile = false) to reduce overhead

Supported Tuning Types

Tuning Type	Support	Notes
Partitioning	Yes	`PARTITION BY` clause at table creation
Sorting	Yes	Sort keys via data model key ordering
Distribution	Yes	`DISTRIBUTED BY HASH` clause at table creation

Table Model

StarRocks uses the Duplicate Key model by default, which is optimal for OLAP workloads. The adapter automatically:

Removes PRIMARY KEY and FOREIGN KEY constraints from source DDL
Adds DUPLICATE KEY(<first_column>) to table definitions
Adds DISTRIBUTED BY HASH(<first_column>) BUCKETS 8 for data distribution

Custom Platform Settings

Additional StarRocks session variables can be applied through tuning configuration:

benchbox run --platform starrocks --benchmark tpch --scale 10.0 \
    --tuning tuned

Troubleshooting

Connection Refused

Error: Failed to connect to StarRocks

Solutions:

Verify StarRocks FE is running and accessible on the configured host and port
Check that port 9030 (MySQL protocol) is open and not blocked by a firewall
For Docker deployments, ensure port mapping is correct (-p 9030:9030)
Test connectivity directly: mysql -h <host> -P 9030 -u root

Missing PyMySQL Dependency

Error: Missing dependencies for starrocks platform: pymysql

Solutions:

Install PyMySQL: uv add pymysql
Or install the StarRocks extra: uv add benchbox --extra starrocks

Schema Creation Failures

Error: Schema creation failed

Solutions:

Check that the StarRocks user has CREATE TABLE and CREATE DATABASE permissions
Verify the target database exists or the user has CREATE DATABASE privilege
Review StarRocks FE logs for detailed error messages
Ensure sufficient disk space and memory on BE nodes

Slow Data Loading

Solutions:

Increase batch_size for larger datasets (default: 1,000 rows per batch)
Consider using Stream Load directly for datasets larger than SF 10
Ensure BE nodes have adequate memory and disk I/O capacity
Check network latency between BenchBox host and StarRocks cluster

Query Timeout

Error: Query execution exceeded timeout

Solutions:

Increase the timeout through the Python API: StarRocksAdapter(max_execution_time=600)
Check StarRocks resource utilization (CPU, memory, disk I/O)
Verify data distribution is balanced across BE nodes
Review the query plan for performance bottlenecks

Search docs and blog

StarRocks Platform

Features

Quick Start

Docker Quick Start

Configuration Options

Data Loading

Loading Process

Stream Load (HTTP API)

Type Mappings

Usage Examples

Basic Benchmarks

Environment Variable Configuration

Custom Database Name

Dry Run (Preview)

Architecture

Adapter Structure

Connection Model

Platform Information

Tuning and Optimization

Automatic Benchmark Configuration

Supported Tuning Types

Table Model

Custom Platform Settings

Troubleshooting

Connection Refused

Missing PyMySQL Dependency

Schema Creation Failures

Slow Data Loading

Query Timeout

See Also