Advanced Docker Swarm
Docker Swarm offers a lot of interesting features that are useful in the Continuous Delivery process. In this section, we will walk through the most important ones.
Rolling updates
Imagine you deploy a new version of your application. You need to update all replicas in the cluster. One option would be to stop the whole Docker Swarm service and to run a new one from the updated Docker image. Such approach, however, causes downtime between the moment when the service is stopped and the moment when the new one is started. In the Continuous Delivery process, downtime is not acceptable, since the deployment can take place after every source code change, which is simply often. Then, how can we provide zero-downtime deployment in a cluster? This is the role of rolling updates.
A rolling update is an automatic method for replacing a service, replica by a replica, in a way that some of the replicas are working all the time. Docker Swarm uses rolling updates by default, and they...
