Native Object Store (NOS) is a capability included with Teradata Vantage that makes it easy for users to explore datasets that have been stored in JSON, comma-separated values (CSV), or Parquet format, located on external object storage like AWS S3 bucket and Azure Blob storage, using standard SQL. Because the data is being processed and analyzed inside the database, Native Object Store can take advantage of the scalability, performance, and workload management that is part of the Vantage platform. No special object storage-side compute infrastructure is required to use Native Object Store. Simply create a Native Object Store table definition within the Database Engine 20, point it at any external object storage you are authorized to access, and within minutes, you can explore the data located in that bucket using all the analytics functionality of Vantage. As a result, Native Object Store is ideally suited for data scientists, analysts, and other business users that want to use object storage on an ad-hoc basis to store interim results, prior versions, and other data sets as part of their analytics workflow.
You can use teradataml to explore this external data made available on Vantage via Native Object Store capability. This section shows how to use teradataml to explore data stored on external object storage with the help of Native Object Store. The first step is to create a foreign table. The foreign table allows the external data to be easily referenced within the Database Engine 20 and makes the data available in a structured relational format which could include complex data types. Once the data is in a relational format, either persistently or virtually, it can be aggregated or joined to other relational tables.
Refer to the Native Object Store documentation for more details about NOS.
The following sections show different ways to explore foreign table data in Vantage via teradataml, based on the actual format of the data. You may refer to the examples in the Native Object Store.