- zero downtime deployment
- gives chance to fix problems and not affect users
- removes requirement to negotiate deployment time
 
	- basic level - know that something is down
- medium level - know that something is going to fail soon
- top level - automatically react (e.g. scale up or down)
- bonus - metrics after deployments
 
	- missing element to verify our ideas
- requires specific approach in application but can (and should) be supported from devops infrastructure
  
		 
		    Made with Slides.com