Why should you care?

DataFlow React app hits Express API which reads orders — pool sizing and timeout settings decide if sale traffic stalls.

See it live — copy this example

Run in MySQL Workbench or the mysql CLI.

-- SQL the API runs (parameterized):
START TRANSACTION;
SELECT stock_qty FROM products WHERE product_id = ? FOR UPDATE;
UPDATE products SET stock_qty = stock_qty - ? WHERE product_id = ?;
INSERT INTO order_items (order_id, product_id, qty) VALUES (?, ?, ?);
COMMIT;

What happened?

• Typical checkout flow in one transaction: lock stock, decrement, insert line.
• Node pool hands connections; each request borrows one connection for txn duration.

Practice next

1. Install mysql2 in Node project; create pool config.
2. Map this SQL in route handler with ? bindings.
3. Set pool size 10; stress test with ab or k6 lightly.
4. Add READ ONLY pool user for GET routes.
5. Set wait_timeout aligned with pool idle timeout.

Remember

Pool connections; do not connect per query anew. Mirror transactional boundaries to business steps. Migrations version schema with app deploys.

DataFlow Express API

PM2 cluster of 4 Node workers shares pool to MySQL on localhost:3306.

Outcome: Mobile app checkout survives moderate concurrent load.