The practices of Continuous Delivery include what needs to be done in order to increase the quality and value of the software. This includes the stability of the application. Reconfiguring and redeploying software does not have to result in any downtime. New features and bugfixes do not have to be shipped exclusively during maintenance windows. Ideally the enterprise software can continuously improve and move forward.

A zero downtime approach involves a certain effort. In order to avoid an application being unavailable, at least one other instance of the software needs to be present at a time. A load balancer or proxy server upfront needs to direct the traffic to an available instance. Blue-green deployments make use this technique:

The application instances including their databases are replicated and proxied by a load balancer. The involved applications typically represent different software versions. Switching the traffic from the blue to the green path and vice versa instantly changes the version, without any downtime. Other forms of blue-green deployments can include scenarios of multiple application instances that are all configured to use the same database instance.

This approach obviously does not have to be realized using some shiny new technology. We have seen blue-green deployments that enable zero-downtime in the past using home-grown solutions. However, modern technologies support these techniques increasing stability, quality, and production-readiness out of the box without much engineering effort required.