---
id: apache-flink
version: "2.3.0"
license: https://www.apache.org/licenses/LICENSE-2.0
license_treatment: permissive
maintenance: active
---

# apache-flink — Apache Flink Python API

License: permissive · Maintenance: active · Downloads: 214.1K/mo

## What it is and what it does

Apache Flink is a distributed processing engine for stateful computations over both bounded (batch) and unbounded (streaming) data. The Python API allows you to build scalable data pipelines using either a high-level Table API (similar to SQL or working with tabular data) or a lower-level DataStream API for fine-grained control over state and time semantics. It runs in common cluster environments and executes computations at in-memory speed across any scale.

The package depends on a substantial stack: py4j for JVM interop, apache-beam, and data-handling libraries like numpy, pandas, and pyarrow. It supports Python 3.9–3.12 and has been in production use since first release in 2020. The Table API suits exploratory data analysis and relational queries, while the DataStream API targets complex stream processing use cases requiring explicit state management.

Use it for:

- Real-time data processing pipelines that ingest, transform, and aggregate streaming events.
- Large-scale batch ETL jobs that read, clean, and load data from distributed storage.
- Machine learning pipelines that prepare and aggregate data for model training at scale.
- Exploratory data analysis on large datasets using SQL-like queries via the Table API.
- Stateful stream processing with windowing, joins, and complex event detection.
- Data quality monitoring and anomaly detection on continuous event streams.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Python API for Apache Flink that builds stateful batch and streaming data processing pipelines, offering both high-level Table API (SQL-like queries) and lower-level DataStream API for complex stream processing.

Yes, if you need a mature, production-grade distributed stream or batch processing framework in Python and can manage the JVM dependency and substantial runtime dependency footprint. Active maintenance, permissive license, and support for Python 3.9–3.12 make it solid for large-scale data pipelines. Not recommended for lightweight, single-machine workloads or if you want to avoid Java runtime overhead.

## Install

pip install apache-flink
uv add apache-flink
poetry add apache-flink

## Installing apache-flink

Before you install: Medium install friction due to 16 runtime dependencies including py4j, apache-beam, numpy, pandas, and pyarrow. Actively maintained with recent release 54 days ago. Wheels available for Python 3.9–3.12 on macOS and Linux.

License in practice: Licensed under Apache License 2.0 (permissive), allowing commercial and private use with minimal restrictions.

Quickstart:

pip install apache-flink

import py4j
# Define and execute streaming or batch job using Table API or DataStream API

Requires Java/JVM runtime; communicates with Flink via py4j. Minimum Python 3.9.

Verify before relying:

- Whether all 16 runtime dependencies are strictly required or conditionally installed based on use case.
- Performance characteristics and scalability limits for typical workloads.
- Whether the DataStream and Table APIs have feature parity or if one is significantly more mature.

## Package facts

- License: https://www.apache.org/licenses/LICENSE-2.0 (permissive)
- Python support: supports_current
- Install friction: medium
- Maintenance: active
- Downloads: 214.1K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

distributed stream processing python, batch and streaming data pipelines, stateful data processing framework, real-time data processing engine, datastream api python, sql-like data queries, large-scale data analysis, stream-processing, distributed-computing, etl

[View on SkillFed](https://skillfed.io/packages/apache-flink) · [View on PyPI](https://pypi.org/project/apache-flink/)