Zeblok Datalake is a high-performance data-store, compatible with the S3 protocol, intended for secure high speed data transfer. As an object store, Zeblok Datalake allows users to read and write large amounts of unstructured data such as photos, videos, log files, spreadsheets, CSV files, etc. Datalake is cloud native and supports GPU-optimized queries through the addition of BlazingDB.
With a robust and simple online interface, API CRUD access, in-built Zeblok CLI magic commands, and Python integrations, users can instantly run complex AI related queries for a variety of datasets, visualize the results, and share/revoke reports with clients.
Follow this brief tutorial to familiarize yourself with Zeblok Datalake.
You can upload files as objects in the Datalake with the simple UI directly from your browser. To get started, go to the Zeblok Computational Dashboard and click on the menu tab that says Datalake.
If you are a new user and have never used Datalake, a new account will be generated for you once you click the Datalake button. As an existing user, clicking the button will automatically log you into Datalake. You should be able to see your objects in the browser.
You can upload files in 3 ways:
- Through the Datalake Browser portal
- Through the Zeblok CLI commands (also known as magic commands)
- Through the Datalake's S3 REST API
Uploading using Datalake Browser portal
To upload through the browser, go to your root folder, which is /datalake// (replace username with yours). Then click the Upload button on the bottom-right corner to select your file to upload.
You can upload using Zeblok's CLI commands in Jupyter notebook. To do that, simply open a new terminal window. Open the terminal by clicking File -> New -> Terminal
You will need to know your username and secret. At the time of writing, we can get this information in Manage Account in the Zeblok Computational Dashboard.
You can login using the Zeblok terminal by saying: zeblok login <username> <secretKey>
Once the terminal is logged in, you can use the zbl commands with these arguments: