This pattern allows for a deduplicated feed of changes from Fauna's transactional data into a secondary data processing pipeline. It uses timestamps from Fauna's database to sort instances by their snapshot order, and then ingests them in that order to track a high-water-mark. This makes it possible to safely resume the data feed after a process restart or server failure. The pattern is designed for full-text search, aggregation, geographic indexes, or any other derived data set that represents an eventually consistent view of the current state of the data. It also handles cases where updates come in faster than the ingester can handle them, and provides guidance on how to define a comfortable polling interval.