Getting Started

Environment Setup

To run TrainDB, the following environment is required. If you have insufficient permission to install, please ask your system administrator. TrainDB includes the TrainDB ML model library as a submodule. The following requirements must be satisfied for operation.

TrainDB

  • Java 11+

  • Maven 3.x

  • SQLite3 (or other DBMS for catalog store, supported by datanucleus)

TrainDB ML Model Library

  • Python 3.8+

  • Python virtual environment manager, such as pyenv, conda (optional)

  • Packages used by ML models, such as pytorch - install requirements.txt

    • Using pip: pip install -r traindb-model/requirements.txt

    • Using conda: conda install –file traindb-model/requirements.txt

Installation

Download

The public repository of TrainDB is https://github.com/traindb-project/traindb. You can download the released file from the address, or you can clone using git as shown below.

$ git clone --recurse-submodules https://github.com/traindb-project/traindb.git

Build

You can build TrainDB using Maven as shown below.

$ cd traindb
$ mvn package

Then, the file traindb-1.5.tar.gz is created in the traindb-assembly/target directory. You can unzip it as shown below.

$ tar xvfz traindb-assembly/target/traindb-1.5.tar.gz