Implementing Domain-Driven Design

Data Replication

Thinking of an in-memory data cache, we may immediately consider the real possibility of losing all or part of our system’s state if the cache fails in some way. It’s a real concern, but far from troublesome when redundancy is built into the Fabric.

Consider the memory cache provided by a Fabric when using a cache-per-Aggregate strategy. In that case the Repository of a given Aggregate type is backed by a dedicated cache. A cache supporting only a single node would be quite vulnerable to failures at a single point. However, a Fabric providing multinode caches with replication would be quite reliable. You can choose the level of redundancy based on the probability of the number of nodes that may fail at any given time, which becomes ...