Star and Snowflake Schemas

What Is a Data Warehouse Schema?

Fact Tables and Dimension Tables

CREATE TABLE fact_sales (
  sale_id      SERIAL PRIMARY KEY,
  date_key     INT NOT NULL,
  product_key  INT NOT NULL,
  customer_key INT NOT NULL,
  store_key    INT NOT NULL,
  quantity     INT NOT NULL,
  revenue      NUMERIC(12, 2) NOT NULL
);

CREATE TABLE dim_product (
  product_key  SERIAL PRIMARY KEY,
  product_name VARCHAR(200),
  category     VARCHAR(100),
  brand        VARCHAR(100),
  unit_price   NUMERIC(10, 2)
);

The Star Schema

-- Star schema: all product info in one flat dimension table
CREATE TABLE dim_product (
  product_key    SERIAL PRIMARY KEY,
  product_name   VARCHAR(200),
  category_name  VARCHAR(100),   -- denormalized
  subcategory    VARCHAR(100),   -- denormalized
  brand_name     VARCHAR(100),   -- denormalized
  brand_country  VARCHAR(100),   -- denormalized
  unit_price     NUMERIC(10, 2)
);

CREATE TABLE dim_date (
  date_key   INT PRIMARY KEY,   -- e.g. 20240315
  full_date  DATE,
  year       INT,
  quarter    INT,
  month      INT,
  month_name VARCHAR(20),
  week       INT,
  day_of_week VARCHAR(10)
);

Star Schema Query

SELECT
  d.year,
  d.quarter,
  p.category_name,
  SUM(f.revenue)   AS total_revenue,
  SUM(f.quantity)  AS units_sold
FROM fact_sales f
JOIN dim_date    d ON d.date_key    = f.date_key
JOIN dim_product p ON p.product_key = f.product_key
WHERE d.year = 2024
GROUP BY d.year, d.quarter, p.category_name
ORDER BY d.quarter, total_revenue DESC;

The Snowflake Schema

-- Snowflake schema: product dimension is normalized
CREATE TABLE dim_brand (
  brand_key     SERIAL PRIMARY KEY,
  brand_name    VARCHAR(100),
  brand_country VARCHAR(100)
);

CREATE TABLE dim_category (
  category_key   SERIAL PRIMARY KEY,
  category_name  VARCHAR(100),
  subcategory    VARCHAR(100)
);

CREATE TABLE dim_product (
  product_key  SERIAL PRIMARY KEY,
  product_name VARCHAR(200),
  category_key INT REFERENCES dim_category(category_key),
  brand_key    INT REFERENCES dim_brand(brand_key),
  unit_price   NUMERIC(10, 2)
);

Snowflake Schema Query

SELECT
  d.year,
  c.category_name,
  b.brand_name,
  SUM(f.revenue) AS total_revenue
FROM fact_sales    f
JOIN dim_date      d ON d.date_key    = f.date_key
JOIN dim_product   p ON p.product_key = f.product_key
JOIN dim_category  c ON c.category_key = p.category_key
JOIN dim_brand     b ON b.brand_key    = p.brand_key
WHERE d.year = 2024
GROUP BY d.year, c.category_name, b.brand_name
ORDER BY total_revenue DESC;

Surrogate Keys vs Natural Keys

-- Surrogate key (product_key) vs natural key (sku)
INSERT INTO dim_product (product_name, category_key, brand_key, unit_price)
VALUES ('Wireless Headphones', 3, 7, 89.99);
-- product_key is assigned by SERIAL -- the natural key (SKU) lives elsewhere

-- Natural key would be:
-- INSERT INTO dim_product (sku, product_name, ...)
-- VALUES ('WH-1000XM5', 'Wireless Headphones', ...);
-- Risky: SKU can be reused or reassigned by the source system

The Date Dimension

-- Populate dim_date for one year using generate_series
INSERT INTO dim_date (date_key, full_date, year, quarter, month, month_name, week, day_of_week)
SELECT
  TO_CHAR(d, 'YYYYMMDD')::INT  AS date_key,
  d                             AS full_date,
  EXTRACT(YEAR    FROM d)::INT  AS year,
  EXTRACT(QUARTER FROM d)::INT  AS quarter,
  EXTRACT(MONTH   FROM d)::INT  AS month,
  TO_CHAR(d, 'Month')           AS month_name,
  EXTRACT(WEEK    FROM d)::INT  AS week,
  TO_CHAR(d, 'Day')             AS day_of_week
FROM generate_series('2024-01-01'::DATE, '2024-12-31'::DATE, '1 day') AS d;

Slowly Changing Dimensions (SCD Type 2)

-- SCD Type 2 customer dimension
CREATE TABLE dim_customer (
  customer_key  SERIAL PRIMARY KEY,
  customer_id   INT NOT NULL,
  customer_name VARCHAR(200),
  city          VARCHAR(100),
  country       VARCHAR(100),
  valid_from    DATE NOT NULL,
  valid_to      DATE,
  is_current    BOOLEAN DEFAULT TRUE
);

-- When a customer moves, close old row and insert new one:
UPDATE dim_customer
   SET valid_to = CURRENT_DATE - 1, is_current = FALSE
 WHERE customer_id = 42 AND is_current = TRUE;

INSERT INTO dim_customer (customer_id, customer_name, city, country, valid_from, is_current)
VALUES (42, 'Alice Muller', 'Berlin', 'Germany', CURRENT_DATE, TRUE);

Star vs Snowflake — Trade-offs

-- Checking how much storage the denormalized category column costs
-- in a large dim_product (star schema) vs a separate dim_category (snowflake)
SELECT
  COUNT(*)                               AS total_products,
  COUNT(DISTINCT category_name)          AS unique_categories,
  pg_size_pretty(
    SUM(pg_column_size(category_name))
  )                                      AS category_storage
FROM dim_product;

Galaxy Schema (Fact Constellation)

CREATE TABLE fact_returns (
  return_id     SERIAL PRIMARY KEY,
  date_key      INT NOT NULL REFERENCES dim_date(date_key),
  product_key   INT NOT NULL REFERENCES dim_product(product_key),
  customer_key  INT NOT NULL,
  quantity      INT NOT NULL,
  refund_amount NUMERIC(12, 2) NOT NULL
);

-- Cross-fact query: net revenue = sales - refunds
SELECT
  d.year,
  d.month,
  SUM(s.revenue)       AS gross_revenue,
  SUM(r.refund_amount) AS total_refunds,
  SUM(s.revenue) - COALESCE(SUM(r.refund_amount), 0) AS net_revenue
FROM dim_date d
LEFT JOIN fact_sales   s ON s.date_key = d.date_key
LEFT JOIN fact_returns r ON r.date_key = d.date_key
WHERE d.year = 2024
GROUP BY d.year, d.month
ORDER BY d.month;

Star vs Snowflake Schema

Lesson Recap