Zarr stores large n-dimensional data

Chunked and scalable. Zarr is a chunked file format for arbitrarily sized n-dimensional arrays, also known as tensors. Storing the data in chunks allows to efficiently cut out small pieces of the data to work with at a time. It also enables writing to the data in parallel without conflicts. The format supports various compression algorithms. These features make Zarr a scalable storage format for even the largest datasets.

Cloud-native. Zarr is designed as a cloud-native format, which enables cost-effective storage and remote access. Data can be stored and managed in centralized storage. Streaming-based access no longer requires copying or transferring of the data. 

Open ecosystem. Zarr libraries for the most popular programming languages (e.g. Python, Java, JavaScript, C/C++, Rust, Julia) are available. A growing list of end-user software also support Zarr. The Zarr specification as well as several library implementations are governed by an open governance model.


How we can help


We are very active in the Zarr community and help to develop the specification as well as library implementations.Since we have implemented full Zarr support in WEBKNOSSOS and Voxelytics, we know everything about it.

    Integrate Zarr support into your software
    Optimizing Zarr data for your use cases
    Distributed computing in the cloud or on your HPC
    Best practice consulting
    Collaborating on specifications and community work
    Workshops on custom topics

Use cases 

Zarr is useful for many scientific domains that deal with large data. Some domains already use Zarr actively.

    Bioimaging (with OME-Zarr)
    Atmosphere, ocean and climate science

Want to work with us?

Schedule an introductory videocall to learn more about our technology and discuss your specific needs with us.