vineyard
An in-memory immutable data manager
What it is and what it does
Vineyard is an in-memory immutable data manager designed to solve the data-sharing bottleneck in distributed big data pipelines. It provides zero-copy sharing of complex data structures (dataframes, tensors, graphs) across different systems via shared memory, eliminating serialization, deserialization, and I/O costs that typically arise when exchanging data between frameworks. The package includes out-of-the-box high-level abstractions for common data types and supports stream pipelining to allow overlapping computation and data transfer.
Vineyard runs as a distributed service with a Python client library that connects to a shared-memory manager. It is a CNCF sandbox project and targets scenarios where data is too large for a single machine and multiple systems need to collaborate on preprocessing, analysis, or training tasks. The 16 runtime dependencies include numpy, pandas, pyarrow, and pyyaml, reflecting its role as a bridge between data processing and machine learning ecosystems.
Use it for:
- Preprocessing large datasets and training models without intermediate file I/O.
- Sharing distributed graphs and tensors across graph analytics engines.
- Building data pipelines where multiple frameworks need to exchange immutable data structures with minimal overhead.
- Reducing memory-copy and I/O costs in multi-stage big data workflows on distributed clusters.
- Integrating numerical computing with machine learning frameworks in a single logical pipeline.
Worth the install?
AI-flagged interpretation of the facts on this page — verify before relying
Vineyard is an in-memory immutable data manager that enables zero-copy sharing of distributed data across big data systems like graph analytics, numerical computing, and machine learning frameworks.
Yes, if you are building distributed big data pipelines that require zero-copy data sharing across multiple frameworks. The permissive Apache License 2.0 and production-stable status support adoption. However, install only if you have a Vineyard server infrastructure in place and are working at scale; it is not a general-purpose data library for single-machine use. The aging maintenance status warrants monitoring for security updates.
Install
vineyard on PyPI
pip
pip install vineyarduv
uv add vineyardpoetry
poetry add vineyardInstalling vineyard
Before you install
Medium install friction due to compiled wheels for multiple Python versions and architectures, but prebuilt binaries are available. Maintenance is aging—last release was 715 days ago, though the repository remains active with recent commits and a stable community.
License in practice
Apache License 2.0 (permissive) allows commercial and private use with minimal restrictions, requiring only license and copyright notice preservation.
Quickstart
pip install vineyard
import vineyard
client = vineyard.connect(vineyard_ipc_socket)
dataset = client.get(vineyard_object_id)
Requires a running Vineyard server instance and an IPC socket path; not a standalone library for local-only use.
Verify before relying
- Whether vineyard-bdist (a listed runtime dependency) is automatically installed or must be configured separately.
- Specific performance characteristics or memory overhead compared to direct shared-memory approaches.
- Compatibility with Windows systems (classifiers list MacOS and POSIX only).
Package facts
| License | Apache License 2.0 (permissive) |
| Python support | not specified |
| Install friction | medium — platform-specific wheel |
| Runtime dependencies | 16 — argcomplete, click, etcd-distro, lazy-import, makefun, numpy, psutil, pyarrow, pyyaml, setuptools, sortedcontainers, treelib, vineyard-bdist, pandas, pickle5, shared-memory38 |
| Maintenance | aging — 715 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 104,367/month — #12,751 on PyPI (30-day window, as of 2026-08-14) |
| Known vulnerabilities | none known (OSV.dev, checked 2026-08-14) |
Evidence: vineyard-0.24.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl; vineyard-0.24.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl; vineyard-0.24.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl; vineyard-0.24.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl; vineyard-0.24.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl; vineyard-0.24.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl; vineyard-0.24.2-cp36-cp36m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl; vineyard-0.24.2-cp36-cp36m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl; vineyard-0.24.2-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl; vineyard-0.24.2-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl; vineyard-0.24.2-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl; vineyard-0.24.2-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl; vineyard-0.24.2-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl; vineyard-0.24.2-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
Tags
More Libraries packages
urllib3 is an HTTP client library that provides…
permissive · top 100 on PyPI
requestsRequests is a Python HTTP library that…
permissive · top 100 on PyPI
pluggyPluggy provides a plugin system that lets you…
permissive · top 100 on PyPI
python-dateutilProvides parsing, arithmetic, and recurrence…
permissive · top 100 on PyPI
sixSix provides utility functions to write Python…
permissive · top 100 on PyPI
pytestpytest is a testing framework that lets you…
permissive · top 100 on PyPI
vineyard-bdistVineyard is an in-memory immutable data manager…
permissive · top 15,000 on PyPI
mooncake-transfer-engine-cuda13Mooncake Transfer Engine provides…
permissive · top 15,000 on PyPI
daskDask is a parallel computing library that…
permissive · top 1,000 on PyPI
mooncake-transfer-engineMooncake Transfer Engine provides…
permissive · top 15,000 on PyPI
apache-flink-librariesBundles Apache Flink's required libraries for…
permissive · top 15,000 on PyPI
onnx-ironnx-ir provides an in-memory intermediate…
permissive · top 5,000 on PyPI
apache-tvm-ffiProvides a stable, minimal C ABI and FFI for…
permissive · top 5,000 on PyPI
litdataLitData optimizes and streams large datasets…
permissive · top 15,000 on PyPI
nvidia-nvshmem-cu12NVSHMEM provides a global address space for GPU…
unclear · top 5,000 on PyPI
pyspark-pandasProvides tools for distributing Pandas…
unclear · top 5,000 on PyPI