Classes
05 - Docker

Docker

Docker is the most well-known containerization technology and has become the de facto standard.

We are going to use Docker because at some point we will have several services interconnected to provide ML services.

Install

If you don't have Docker installed, go to this link and install it.

Start a service

Let's get a PostgreSQL database service up. First, let's do it via the command line:

$ docker run -d --name postgres-mlops -p 5100:5432 -e POSTGRES_PASSWORD=abc123 -v pg-data-mlops:/var/lib/postgresql/data postgres:16-alpine

Question 1

Answer

postgres-mlops

Question 2

Answer

5100

Question 3

Answer

pg-data-mlops is the name of the Docker volume that is being created to persist the PostgreSQL data.

This means that the data will be stored outside the container's filesystem, allowing it to persist even if the container is removed.

The data will be accessible at the path /var/lib/postgresql/data inside the container.

Check created volumes with:

$ docker volume ls

Question 4

DB Connection

Now let's connect to the PostgreSQL database from DBeaver.

Open DBeaver and create a new database connection. As credentials, you can use:

HOST: "localhost"
USERNAME: "postgres"
PASSWORD: "abc123"
DB: "postgres"
PORT: 5100

Tip! 1

When creating the connection, on the second tab check the option "Show all databases"

Thus, if you create new databases, they will be displayed automatically.

Attention!

Depending on your version of DBeaver, the "Show all databases" option may be available in the "Main" tab, next to the "Database" field.

DDL - Data Definition Language

Create the database used in the previous class. Here is the DDL from the item_sale table:

Important!

If the schemas and tables were created correctly, this change is a mere deploy information.

CREATE SCHEMA IF NOT EXISTS sales;
CREATE SCHEMA IF NOT EXISTS sales_analytics;

DROP TABLE IF EXISTS sales.item_sale;
CREATE TABLE sales.item_sale (
  id SERIAL PRIMARY KEY,
  store_id integer,
  client_id integer,
  product_id integer,
  date_sale DATE,
  price decimal(10,2)
);


DROP INDEX IF EXISTS item_sale_store_id_idx;
CREATE INDEX IF NOT EXISTS item_sale_store_id_idx
  ON sales.item_sale (store_id,date_sale);

Inserts in this table can be done with:

INSERT INTO sales.item_sale
    (store_id, client_id, product_id, date_sale, price)
VALUES
    (5000, 148331, 1715, '2022-01-01', 362.16),
    (5000, 203423, 1921, '2022-01-01', 334.23),
    (5000, 192737, 1667, '2022-01-01', 351.77),
    (5005, 369515, 2131, '2023-08-22', 877.95);

Question 5

Question 6

Attention!

If the schemas and tables were created correctly, this change is a mere deploy information.

So, we will have changes in the .env and nothing in the code!