Introducing the StatefulSet object

You may wonder why running stateful workloads in the distributed cloud is generally considered harder than stateless ones. In classic three-tier applications, all the states would be stored in a database (data tier or persistence layer) and there would be nothing special about it. For SQL servers, you would usually add a failover setup with data replication, and in case you require superior performance, you would scale vertically by simply purchasing better hardware for hosting. Then, at some point, you might think about clustered SQL solutions, introducing data sharding (horizontal data partitions). But still, from the perspective of a web server running your application, the database would be just a single connection string to read and write the data. The database would be responsible for persisting a mutable state.

Important note

Remember that every application as a whole is, in some way, stateful unless it serves static content only or just...