Course of Infrastructure Design: Transactions

💸 First lesson available without registration 💸

When we have a clean architecture, we typically have one repository for each use case. This implies that usually in each repository we only have one table, so we often do not need transactions.

The downside is that there’s a catch, and a big one. After each mutation (insertion or modification), we want to publish a domain event. And what happens if our queue system goes down?

This is where transactions come in handy. And although it may seem like a trivial decision, where we place it will determine how maintainable our code ends up being.

In this course, we will cover:

• 👀 Where to make transactions: In the repository, in the use case, wrapping everything?
• 🔙 Rollback of non-transactional operations: How can we perform the inverse operation?
• 🥊 Comparative of transactions: The final decision