Features

Purpose of this project: Users can integrate No-SQL data sources into their Exasol database as a Virtual Schema

Mapping to Fixed Data Types

The Virtual Schema adapter converts the semi-structured document data into a relational data format.

That has two main challenges:

Different data types in each row

The virtual schema adapter dynamically converts the source data to the target data type.

Rationale:

In semi-structured data, each document can have a different structure. It’s also possible that a property has different types between different documents.

Handle nested lists

The virtual schema adapter can convert nested lists from documents so that they can be processed in Exasol.

Rationale:

The relational model of Exasol does not support nested lists.

Filter Push-Down

feat~filer-pushdown~1

The Virtual Schema adapter pushes as many predicates from the WHERE clause as possible down to the data source.

Needs: genericdsn, dialectdsn

Support for Different Data Sources

We offer Virtual Schema adapter for accessing the following types of No-SQL data:

• DynamoDB
• Files
  • with file types:
    • JSON
    • JSON-lines
    • Parquet
  • with storage backends
    • from AWS S3
    • from Google Cloud Storage
    • from BucketFS
    • from Azure Blob Storage
    • from Azure Data Lake Gen2 Storage

Fast Loading for Different Load Patterns

The Virtual Schema performs well for the following load patterns:

• Many small files
• One / few big files
• One / few small files