---
id: distributed
version: "2026.7.1"
license: BSD-3-Clause
license_treatment: permissive
maintenance: active
---

# distributed — Distributed scheduler for Dask

License: permissive · Maintenance: active · Downloads: 8.4M/mo

## What it is and what it does

Distributed is a scheduler and runtime that coordinates parallel and distributed computation for Dask. It manages task scheduling, data movement, and fault tolerance across a cluster of workers, allowing you to run computations on multiple machines or cores as if they were a single pool. The package handles the low-level coordination—worker lifecycle, task graph execution, communication—so you focus on expressing your computation in Dask.

You typically use it by creating a Client that connects to a cluster (local or remote), then submitting Dask graphs or collections through that client. It depends on Tornado for async networking, Cloudpickle for serialization, and several utility libraries (toolz, sortedcontainers, locket, msgpack) to manage scheduling state and inter-process communication. The package is production-stable and actively maintained.

Use it for:

- Scale data processing pipelines across a cluster without rewriting code for distributed execution.
- Run machine learning training or inference jobs in parallel across multiple GPUs or nodes.
- Execute long-running batch computations that would timeout or exhaust memory on a single machine.
- Coordinate complex workflows with dependencies between tasks across a heterogeneous cluster.
- Monitor and debug distributed workloads through the scheduler's built-in diagnostics and web dashboard.

## Worth the install?

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

Distributed provides a scheduler and runtime for parallel and distributed computation using Dask, enabling you to scale Python workloads across multiple machines or cores.

Yes. Distributed is production-stable, actively maintained, and widely used for scaling Dask workloads. Install friction is low, dependencies are well-established, and there are no known vulnerabilities. Choose it if you need to parallelize or distribute computation beyond a single machine; if you only need local parallelism, Dask alone may suffice.

## Install

pip install distributed
uv add distributed
poetry add distributed

## Installing distributed

Before you install: Low install friction with a pure-Python wheel. Active maintenance with a recent release (31 days ago) and ongoing commits. Supports current Python versions (3.10–3.14).

License in practice: BSD-3-Clause is permissive; you may use, modify, and distribute this package freely in commercial and private projects with minimal restrictions.

Quickstart:

pip install distributed

from distributed import Client

client = Client()
result = client.submit(lambda x: x + 1, 1).result()

Requires Python 3.10 or later. Typically used with Dask; standalone use is possible but limited.

Verify before relying:

- Whether the package works reliably on all supported Python versions (3.10–3.14) in production environments.
- Performance characteristics and scalability limits for different cluster sizes and workload types.

## Package facts

- License: BSD-3-Clause (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 8.4M/month (top 5,000 on PyPI)
- Known vulnerabilities: none known

## Tags

distributed task scheduler, parallel computation framework, dask distributed computing, multi-machine workload scaling, cluster job scheduler python, parallel-computing, cluster-scheduling, dask-ecosystem

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