Beetle (in development)

High performance ETL implemented in golang concurrent fan-out pattern and scylla node clusters.

ETL pipeline

Messages are retrieved from sources defined in conf.json by a producer that downloads and parses files into data ready to be consumed
The producer sends the messages data to a pool of consumer workers that will validate, transform and load the messages into a database
Consumer workers clean the messages marking any that does not pass a filter with dirty_fields: true
Consumer workers send the failed messages to the failed_imports table and successful messages to the messages table (todo)

Setting Up Local Development Environment

Install Dependencies

Install golang: brew install golang
Install Docker: brew install homebrew/cask/docker

Start the scylla cluster

I created a docker compose file which will take the files in scylla_config and create 3 containerized scylla nodes networked with each other. The host machine can start a scylla session at 127.0.0.1:9042

Start scylla node cluster: docker compose up

Configure and run beetle

Edit the conf.json file to add Urls to retrive messages from: cp example_conf.json conf.json
Run bettle: go run .
If you want to compile a binary and run: mkdir -p bin && go build -o bin/beetle && bin/beetle

This is not necessary to run, but if you want to generate a go file to see all scylla's table metadata (columns, parkeys, sortkeys) do this:

Generating table metadata (columns, partkeys, sortkeys) from scylla

Install gocqlx's schemagen: go get -u "github.com/scylladb/gocqlx/v2/cmd/schemagen"
If you have java, make sure you have $GOBIN defined in your path (location to your go bin directory) or you may run java's schemagen
generate schema go file: $GOBIN/schemagen -cluster="127.0.0.1:9042" -keyspace="messages" -output="schema" -pkgname="schema"

jeraldrich / beetle