- trade scalability for latency
- inappropriate for very low-latency notifications
- seconds (or minutes or hours) between events being produced and consumed
- Suited for many consumers, guaranteed delivery, but latency-tolerant
- Clients poll the service feed, use caching to reduce workload. Clients responsible for retrieving (inverse of many alternatives)
- Messages will never arrive out-of-order.
- Easy to replay entire history.
- This linking and caching strategy trades efficiency for generalization.
- Read more
- REST in Practice, chapter 7: The Atom syndication format