Nowadays, Batching is not the first choice when we think about getting data into our systems. Streaming is becoming the default mechanism for these kind of tasks. However, streaming requires fine grain architecture to maintain everything inmutable, a important role played by the CQRS pattern. This time we will see how streaming technologies, such as Apache Kafka, integrate well CQRS and how we segregate our system writes from our reads. We will explain how we can create diverse read models from a single inmutable and consistent model while maintaining clients of each model separated from each other without compromising performance.