zarr-developers/zarr-python

Health v2 · maintenance only

Adoption not part of the score

2044 stars · 445 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Zarr is a Python library implementing the Zarr storage format for chunked, compressed, N-dimensional arrays with NumPy-compatible dtypes. It supports storing arrays in memory, on disk, in zip files, or on object stores like S3, with concurrent reads and writes across threads and processes.

Use cases

• store large numpy arrays on S3 in chunked compressed format
• read and write n-dimensional arrays concurrently from multiple processes
• organize scientific datasets into hierarchical groups of arrays
• stream chunks of a huge array without loading it all into memory
• compress array data on disk for efficient storage
• use zarr arrays as backing store for distributed computing workflows

When to choose

• you need NumPy-compatible arrays larger than memory with chunked access
• you want cloud/object-storage-friendly scientific array storage
• you need concurrent parallel reads or writes to shared arrays
• you work with large imaging, geospatial, or simulation datasets

When to avoid

• you only need small in-memory arrays that plain NumPy handles fine
• you need relational or tabular data storage rather than arrays
• you require a single-file format like HDF5 for portability

Facets

library · maturity stable

serialization compression file-system data-science data-science big-data cloud-computing developer-tools python cross-platform cloud ndarray chunked-arrays numpy object-storage parallel-computing scientific-computing storage-format

Member repositories

For agents

markdown · JSON · MCP: product_card(name="zarr-developers/zarr-python")