1.8 KiB

Raw Blame History

Introduction

This project implements an abstraction of objects that can have access to a variety of data stores, implementing read/write with a simple and expressive interface. This abstraction works with NoSQL, SQL and Cloud data stores and leverages pandas.

Why Use Data-Transport ?

Data transport is a simple framework that:

easy to install & modify (open-source)
enables access to multiple database technologies (pandas, SQLAlchemy)
enables notebook sharing without exposing database credential.
supports pre/post processing specifications (pipeline)

Installation

Within the virtual environment perform the following (the following will install everything):

pip install data-transport[all]@git+https://github.com/lnyemba/data-transport.git

Options to install components in square brackets are nosql; cloud; other and warehouse

pip install data-transport[nosql,cloud,other, warehouse,all]@git+https://github.com/lnyemba/data-transport.git

The components available:

0. sql          by default netezza; mysql; postgresql; duckdb; sqlite3; sqlserver
1. nosql        mongodb/ferretdb; couchdb
2. cloud        s3; bigquery; databricks
3. other        files; http; rabbitmq
4. warehouse    apache drill; apache iceberg

Additional features

- Reads are separated from writes to avoid accidental writes.
- Streaming (for large volumes of data) by specifying chunksize
- CLI interface to add to registry, run ETL
- Implements best-pracices for collaborative environments like apache zeppelin; jupyterhub; SageMaker; ...

Learn More

We have available notebooks with sample code to read/write against mongodb, couchdb, Netezza, PostgreSQL, Google Bigquery, Databricks, Microsoft SQL Server, MySQL ... Visit data-transport homepage