ParadeDB Platform

CLI name: paradedb - use benchbox run --platform paradedb

ParadeDB is a PostgreSQL extension for hybrid search and analytics workloads. Its pg_analytics extension adds Elasticsearch-compatible BM25 full-text search over ordinary PostgreSQL heap tables. BenchBox provides first-class ParadeDB support for benchmarking analytics workloads on ParadeDB-backed PostgreSQL.

Features

Heap tables - Benchmark tables stay ordinary PostgreSQL heap tables loaded through the inherited COPY path
Extension gating - Connection setup verifies (or creates) pg_analytics with an actionable install error otherwise
Version provenance - The installed pg_analytics version is recorded in platform info
PostgreSQL compatible - Standard PostgreSQL wire protocol, dialect, and COPY loading

Quick Start

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

benchbox run --platform paradedb --benchmark tpch --scale 1.0 \
  --platform-option host=paradedb.example.com \
  --platform-option password=secret

Installation

Python Dependencies

ParadeDB uses the same Python driver as PostgreSQL (psycopg v3):

pip install "benchbox[postgresql]"

Server Requirements

ParadeDB must be installed on the PostgreSQL server. The recommended approach is Docker.

Docker (recommended):

docker run -d --name paradedb \
  -e POSTGRES_PASSWORD=benchbox \
  -p 5432:5432 \
  paradedb/paradedb:latest

psql -h localhost -U postgres -c "CREATE EXTENSION pg_analytics;"

See the ParadeDB GitHub repository for additional installation methods.

Configuration Options

Option	Default	Description
`host`	`localhost`	PostgreSQL server hostname
`port`	`5432`	PostgreSQL server port
`database`	auto-generated	Database name
`username`	`postgres`	PostgreSQL username
`password`	(none)	PostgreSQL password
`schema`	`public`	PostgreSQL schema name

Recommended Benchmarks

TPC-H - Standard OLAP benchmark on ParadeDB-backed heap tables
TPC-DS - Complex OLAP; same heap-table path as PostgreSQL