←Back to Documentation
Hello World Tutorial
Multi-model queries in action - see entropyDB's power in one example
What You'll Build
In this tutorial, you'll build a simple application that demonstrates entropyDB's multi-model capabilities. You'll work with relational data, JSON documents, time-series events, and vector embeddings—all in a single database.
💡 What You'll Learn
- Creating tables with multiple data types
- Querying JSON documents
- Working with time-series data
- Performing vector similarity search
- Cross-model joins and queries
Step 1: Create the Schema
Let's create a simple e-commerce schema that uses all data models:
-- Relational: Users table CREATE TABLE users ( id SERIAL PRIMARY KEY, username TEXT NOT NULL UNIQUE, email TEXT NOT NULL, created_at TIMESTAMP DEFAULT NOW() ); -- Document: Product catalog with flexible schema CREATE TABLE products ( id SERIAL PRIMARY KEY, name TEXT NOT NULL, price DECIMAL(10,2), metadata JSONB, -- Flexible product attributes embedding VECTOR(768), -- For semantic search created_at TIMESTAMP DEFAULT NOW() ); -- Time-Series: User activity events CREATE TABLE events ( timestamp TIMESTAMP DEFAULT NOW(), user_id INT REFERENCES users(id), event_type TEXT, properties JSONB ); -- Create indexes CREATE INDEX idx_products_metadata ON products USING GIN(metadata); CREATE INDEX idx_products_embedding ON products USING hnsw(embedding vector_cosine_ops); CREATE INDEX idx_events_timestamp ON events(timestamp DESC);
Step 2: Insert Sample Data
Add Users
INSERT INTO users (username, email) VALUES
('alice', 'alice@example.com'),
('bob', 'bob@example.com'),
('charlie', 'charlie@example.com');Add Products with JSON Metadata
INSERT INTO products (name, price, metadata) VALUES
('Laptop', 999.99, '{
"brand": "TechCorp",
"specs": {"cpu": "Intel i7", "ram": "16GB", "storage": "512GB SSD"},
"tags": ["electronics", "computers", "featured"]
}'::jsonb),
('Wireless Mouse', 29.99, '{
"brand": "TechCorp",
"specs": {"dpi": "4000", "wireless": true},
"tags": ["electronics", "accessories"]
}'::jsonb),
('USB-C Cable', 12.99, '{
"brand": "CableCo",
"specs": {"length": "2m", "speed": "USB 3.1"},
"tags": ["accessories", "cables"]
}'::jsonb);Add User Events
INSERT INTO events (user_id, event_type, properties) VALUES
(1, 'page_view', '{"page": "product", "product_id": 1}'::jsonb),
(1, 'add_to_cart', '{"product_id": 1, "quantity": 1}'::jsonb),
(2, 'page_view', '{"page": "product", "product_id": 2}'::jsonb),
(2, 'purchase', '{"product_id": 2, "amount": 29.99}'::jsonb),
(3, 'search', '{"query": "laptop", "results": 5}'::jsonb);Step 3: Multi-Model Queries
Query 1: JSON Document Search
Find all electronics with specific specs:
-- Find products with "electronics" tag
SELECT
name,
price,
metadata->>'brand' as brand,
metadata->'tags' as tags
FROM products
WHERE metadata @> '{"tags": ["electronics"]}';Query 2: Time-Series Analysis
Analyze user activity over time:
-- Events by hour with user details
SELECT
DATE_TRUNC('hour', e.timestamp) as hour,
e.event_type,
COUNT(*) as event_count,
array_agg(DISTINCT u.username) as users
FROM events e
JOIN users u ON e.user_id = u.id
GROUP BY hour, e.event_type
ORDER BY hour DESC;Query 3: Complex Join Across Models
Find user purchase behavior with product details:
-- User activity report with product info
SELECT
u.username,
u.email,
e.event_type,
p.name as product_name,
p.metadata->>'brand' as brand,
e.properties->>'amount' as amount,
e.timestamp
FROM users u
JOIN events e ON u.id = e.user_id
LEFT JOIN products p ON (e.properties->>'product_id')::int = p.id
WHERE e.event_type IN ('add_to_cart', 'purchase')
ORDER BY e.timestamp DESC;Query 4: Aggregations with JSON
Product statistics by brand:
-- Brand analytics SELECT metadata->>'brand' as brand, COUNT(*) as product_count, AVG(price) as avg_price, array_agg(DISTINCT jsonb_array_elements_text(metadata->'tags')) as all_tags FROM products GROUP BY metadata->>'brand' ORDER BY product_count DESC;
Step 4: Add Vector Search
Let's add semantic search capabilities using vector embeddings:
-- Update products with sample embeddings -- In production, you'd generate these from actual product descriptions UPDATE products SET embedding = random_vector(768) -- entropyDB helper function WHERE id = 1; -- Semantic search for similar products SELECT p1.name, p1.price, p1.metadata->>'brand' as brand, 1 - (p1.embedding <=> p2.embedding) as similarity FROM products p1 CROSS JOIN products p2 WHERE p2.id = 1 -- Find products similar to product 1 AND p1.id != p2.id ORDER BY p1.embedding <=> p2.embedding LIMIT 5;
💡 Pro Tip
In a real application, you'd generate embeddings using OpenAI, Cohere, or other embedding models based on product descriptions.
Step 5: Multi-Model Transactions
entropyDB ensures ACID guarantees across all data models:
-- Complete purchase transaction across multiple models
BEGIN;
-- Update user data (relational)
UPDATE users
SET metadata = jsonb_set(
COALESCE(metadata, '{}'::jsonb),
'{last_purchase}',
to_jsonb(NOW())
)
WHERE id = 1;
-- Record event (time-series)
INSERT INTO events (user_id, event_type, properties)
VALUES (1, 'purchase', '{"product_id": 1, "amount": 999.99}'::jsonb);
-- Update product analytics (document)
UPDATE products
SET metadata = jsonb_set(
metadata,
'{sales_count}',
to_jsonb(COALESCE((metadata->>'sales_count')::int, 0) + 1)
)
WHERE id = 1;
COMMIT;
-- Everything succeeds or nothing does!Complete Example Output
Here's what a complete query result looks like:
username | event_type | product_name | brand | amount | timestamp ---------|--------------|--------------|----------|---------|------------------------- bob | purchase | Mouse | TechCorp | 29.99 | 2025-11-17 10:30:15 alice | add_to_cart | Laptop | TechCorp | null | 2025-11-17 10:25:42 charlie | search | null | null | null | 2025-11-17 10:20:10 (3 rows)
Next Steps
Now that you've seen entropyDB's multi-model capabilities, dive deeper into specific features: