Gin - A Distributed Data Processing Framework

Gin is a Distributed Data Processing Framework implemented in Rust. It offers a DataFrame interface for processing Parquet files. Gin Library provides an API that allows reading Parquet files from S3 into a Dataframe and lazily run the transformations on it which are then triggered by actions.

Gin currently supports:

• Transformations:
  • select - Select columns.
  • filter - Filter rows.
• Actions:
  • width - Get the number of columns.
  • count - Count the number of rows.
  • sum - Sum of the values of the specified column.

Prerequisites

$ apt install -y protobuf-compiler

Installation

$ git clone https://github.com/malvag/gin-framework.git
$ cd gin-framework
$ cargo build --release

Run Gin

Start a Gin Cluster, specifying the number of executors.

$ ./gin-context start 4

Run the test client.

Note: You should change the S3 Credentials first and set up the bucket. We used MinIO for simplicity.

$ cargo run --release

Stop the cluster.

$ ./gin-context stop

Run MinIO

Run MinIO localy using docker.

$ docker run -p 9000:9000 -p 9090:9090 --name minio \
    -v ~/rust_c/s3-read/minio/data:/data \
    -e "MINIO_ROOT_USER=minioadmin" \
    -e "MINIO_ROOT_PASSWORD=minioadmin" \
    quay.io/minio/minio server /data --console-address ":9090"

As dataset, we used the yellow-taxi parquet file.

Contributors

• Alexandra Alexandridi
• Evangelos Maliaroudakis
• Ioannis Melidonis