We'll start the exploration of stateful services in a Swarm cluster by taking a look at what would happen if we deploy them as any other service.
A good example is Jenkins. Every job we create is an XML file. Every plugin we install is an HPI file. Every configuration change is stored as XML. You get the picture. Everything we do in Jenkins ends up being a file. All those files form its state. Without it, Jenkins would not be able to operate. Jenkins is also a good example of the problems we have with legacy applications. If we were to design it today, it would probably use a database to store its state. That would allow us to scale it since all instances would share the same state by being connected to the same database. There are quite a few other design choices we would probably make if we were to design it today from scratch. Being legacy is not necessarily a bad thing. Sure, having the experience we have today would help us avoid some...