Cloud Native Development Patterns and Best Practices

Database schema

Database versioning in a cloud-native system is very different. Traditionally, we would perform a database conversion. However, conversions often require downtime. Furthermore, a batch-oriented conversion may not be reasonable in a cloud-native system because of the volume of data. Alternatively, an A/B feature flag may necessitate running concurrent versions.

Cloud-native databases are often schemaless. This eliminates a great deal of complexity that is traditionally required to deploy a new database or change a database schema. However, the data structure stored in a cloud-native database does have an implicit schema. Therefore, the code must be implemented to handle multiple versions of the implicit schema. Each row could ...