The Old Way vs. The New Way
In the early days of software operations, updating an application meant scheduling a "maintenance window." You would put up a "Site Under Construction" page on Saturday night, shut down the servers, install the new version, cross your fingers, and hope everything came back online by Sunday morning. If something went wrong, you were debugging at 3 AM with angry stakeholders on the phone.
In today's 24/7 world, that model is unacceptable. A major e-commerce platform losing availability for two hours costs millions in revenue. A SaaS product going dark during business hours loses customer trust permanently. Users expect services to be available continuously — even while those services are being updated.
The Goal: Zero Downtime Deployments
Deployment strategies are techniques for releasing new versions of software into production without interrupting service for users. They are not just about avoiding downtime — they are about managing risk. Every deployment is a bet: you are betting that the new version is better than the old one. Deployment strategies let you control the size of that bet.
They help answer three critical operational questions:
- Availability: How do we update the application without users noticing any interruption?
- Rollback speed: If the new version is broken, how quickly can we revert to safety?
- Risk exposure: How many users are affected if the new version has a bug?
The four strategies covered in this lesson — Rolling Updates, Blue/Green, Canary, and Shadow — each make different trade-offs across these three dimensions. Understanding those trade-offs is what allows you to choose the right strategy for each situation.