This is a framework for working with SQL transactions in a distributed network. The framework allows you to compose SQL transactions of multiple queries on different nodes and execute them later.
To ensure the reliability of the transactions a ledger using blockchain is used to keep track on ongoing transactions and ensure queries within the transactions are consistent.
The core library containing the DBTModule to include in your Ratpack application.
The DBT module can be added like so:
bindings {
module (DBTModule) { config ->
config.ledger.remoteUrl = '<ledger host>:<ledger port>/ledger'
config.executor.remoteUrl = '<executor host>:<executor port>/executor'
}
}
The test module contains all the integration tests for the module.
There are two projects available for building the executor and the ledger as docker images.
- docker-ledger - Run the ledger as a docker image
- docker-executor - Run the executor as a docker image
For more information how to run and configure them see their corresponding READMEs.
The project also comes with a docker-compose.yml file for running the docker images.
To run the images first you need to buid them and publish them to your local docker registry.
You can do that by running:
./gradlew dbt-executor:distDocker
./gradlew dbt-ledger:distDocker
After you have generated the Docker images you can run them both by running docker-compose up
in the root folder.
The ledger will be available at http://localhost:5050/ledger and the executor will be available at http://localhost:5051/executor
Once you have the executor and the ledger configured and running, you can start making transactions to the database the executor is connected to.
By default the executor will set up an in-memory database without any tables, so the first transaction you want to make is to create the table you want.
The preferred way would be to use one the libraries provided to communicate with the ledger, but you can also try out the functionality by posting JSON directly to the ledger.
Here is an example transaction you can POST to the ledger. In this example we post a completed transaction which will be executed immediatly after registering with the ledger. If you want to postpone execution set to completed property to false.
POST http://localhost:5050/ledger
{
"completed" : true,
"queries": [
{ "query" : "CREATE TABLE foo(bar INT)" },
{ "query" : "INSERT INTO foo (bar) VALUES (5)" },
{ "query" : "SELECT * FROM foo" }
]
}
The response you will get back looks something like this
{
"id": "5db5f0d506994a54b6482f587a953c43",
"queries": [
{
"query": "CREATE TABLE foo(bar INT)",
"id": "d1e2a4d4d5b3233ef07bc52ad935f459da26e5917c542c499541538156007d94",
"parent": "5db5f0d506994a54b6482f587a953c43",
"timeStamp": 1528306534048,
"result": null,
"resultError": null
},
{
"query": "INSERT INTO foo (bar) VALUES (5)",
"id": "c3df3ea4402f4b4aa0ffec1fd728fe00a013106a46719846362b9629b5a3e4f6",
"parent": "d1e2a4d4d5b3233ef07bc52ad935f459da26e5917c542c499541538156007d94",
"timeStamp": 1528306534050,
"result": null,
"resultError": null
},
{
"query": "SELECT * FROM foo",
"id": "ef59de620262b5a41c2944f9539c27dd375e1123d6c43b1251f407ae6c34ec59",
"parent": "c3df3ea4402f4b4aa0ffec1fd728fe00a013106a46719846362b9629b5a3e4f6",
"timeStamp": 1528306534050,
"result": {
"BAR": [
5
]
},
"resultError": null
}
],
"completed": true,
"executed": true,
"rolledback": false
}
As you can see, the queries were made and the results returned successfully.
The ledger will automatically create the necessery hashes for the blockchain.
If you do a GET request to the ledger URL (http://localhost:5050/ledger) you can see all transactions and their state.