Deployment

This section covers everything you need to deploy an Apache Pinot cluster, from bootstrapping your first ZooKeeper-connected cluster through configuring tables, ingestion pipelines, and component-level settings. Use these guides to stand up Pinot in any environment -- bare metal, VMs, or containers -- before moving on to production hardening.

When to use these guides

• You are deploying Pinot for the first time outside of the quickstart.

• You need to start individual Pinot components (controller, broker, server) with custom configuration files.

• You are setting up tables, schemas, and ingestion jobs against a running cluster.

• You want to optimize your configuration using the recommendation engine or decouple the controller from the real-time data path.

Prerequisites

• Java 11 or later installed (or Docker if using container-based deployment).

• A running Apache ZooKeeper ensemble (three nodes recommended for production).

• Access to a deep store (local filesystem for development; S3, GCS, HDFS, or ADLS for production).

• The Pinot binary distribution downloaded from pinot.apache.org/download or the official Docker image.

Child pages

Cluster setup

Set up a Pinot cluster walks through creating a cluster namespace and starting controller, broker, and server instances.

Advanced Pinot setup

Advanced Pinot Setup provides step-by-step instructions for starting each Pinot component individually using Docker or launcher scripts, configuring components with custom config files, creating batch and streaming tables, and loading data.

Server startup status checkers

Server Startup Status Checkers explains readiness and liveness health checks, static consumption wait, offset-based and freshness-based segment checkers, and recommended configurations for QA and production environments.

Table setup

Set up a table links to table creation instructions including schema design, table configuration, and index settings.

Ingestion setup

Set up ingestion covers creating segments and running batch data ingestion jobs to load data into your tables.

Decoupling controller from the data path

Decoupling Controller from the Data Path explains how to bypass the controller during real-time segment completion by uploading completed segments directly to deep store, and how to enable peer download for failure recovery.

Command-line interface (CLI)

CLI Reference documents every pinot-admin.sh command -- from AddSchema and AddTable through RebalanceTable, LaunchDataIngestionJob, and cluster validation utilities.

Configuration recommendation engine

Configuration Recommendation Engine describes the rule-based engine accessible via the controller REST API that recommends optimal index, partitioning, segment size, and real-time provisioning settings based on your schema, query patterns, and workload characteristics.

Next step

Once your cluster is deployed and serving data, harden it for production use. See Production Guides for capacity planning, health checks, graceful operations, and disaster recovery.