Opening Scenario
Your team owns a real-time analytics dashboard: hourly active users, conversion funnels, and revenue trends, updated every few seconds. The CTO walks in with three asks: dashboards must update in seconds, last-year-to-date numbers must reconcile to the penny with finance, and analysts must be able to define new metrics and backfill them across two years of history. The architect has whiteboarded two diagrams labeled "Lambda" and "Kappa." Half the room wants Lambda ("the speed layer for fast, the batch layer for correct"), the other half wants Kappa ("just stream everything, replay when needed"). You have to pick one — and live with it. Before you commit, you need to understand what these architectures actually are, what each is paying for, and which fits the real workload.
Question Progression
1. What is the Lambda architecture, concretely?
Question. Where does Lambda's "two layers" idea come from, and what does each layer actually do?
In the full tutorial
- When to Reach for This