Any system must start with data to process. On serverless platforms, there are multiple choices for streaming systems. Azure, Google Compute, and AWS all offer some form of streaming systems. I mentioned these in Chapter 6, Asynchronous Processing with the Messaging Pattern, when discussing the differences between queues and streams:
- Azure: Event Hubs
- AWS: Kinesis
- Google Compute Cloud: Cloud Dataflow
It's worth briefly touching on the topic of queues versus streams again. As mentioned in Chapter 6, Asynchronous Processing with the Messaging Pattern, one of the main differentiators is that queues are primarily designed for once-only processing. That is, once a message is pulled from a queue, no other consumer will ...