TileDB Embedded

The Challenge

✘A sea of files and legacy formats
✘Formats not designed for cloud
✘Data conversion before analysis
✘Data updates hard to handle

The Solution

Multi-dimensional arrays
Cloud-native format
Interoperability with zero-copy
Built-in data versioning

Arrays, not files

Model any complex multi-dimensional data as arrays. Dataframes, genomic variants, satellite images or time-series, can all be efficiently represented as dense or sparse arrays. TileDB implements a universal array format that captures all data science applications.

Optimized for the cloud

TileDB is built with cloud object store challenges in mind and delivers superior performance via an optimized protocol and parallel IO. TileDB works on AWS S3, Google Cloud Storage and Azure Blob Storage.

Data science ecosystem

TileDB offers numerous APIs (C, C++, Python, R, Java, Go) and integrations (PrestoDB, MariaDB, Dask, Spark, PDAL, GDAL), eliminating data conversions with zero-copying wherever possible.

See all API Integrations

>>> import tiledb
>>> array = tiledb.open("s3://tiledb-inc-demo-data/2.0/example")
>>> array.shape
(100, 100, 30)

>>> array.dtype
dtype("float64")

>>> np.mean(array[:,:,1:10])
0.49943713803540135

Data updates and time traveling

TileDB supports data versioning with rapid updates and time traveling, all built into its cloud-native array data format and storage engine.

GitHub

Join the growing TileDB open source community on Github and shape the future of data science.

Go to code

Documentation

TileDB Embedded is an exciting technology. There is a lot more to explore, visit our docs to get started.

View docs

Forum

Your use cases are shaping TileDB as the universal data engine. Join the discussions on the TileDB forum.

Visit forum

TileDB Embedded

Store any data as multi-dimensional arrays in a cloud-native, open-source format

Use TileDB for free with your own compute from various data science tools