A pull query answers a one-time state question: it runs on demand, retrieves current data, returns the result, and terminates. Unlike a push query, it does not continuously process an unbounded stream or emit later updates. section 8.4.1
Choose the lookup path according to where the current state resides:
- Querying a stream or table scans the underlying Kafka topic, changelog, or original stream for the most recent matching record. section 8.4.1
- Querying a materialized view reads a continuously updated, locally stored view backed by RocksDB and replicated through Kafka, avoiding a topic scan. section 8.4.1
Current order state
A request such as “What is the status of order 123?” needs the state at query time, not a continuously running application. A high-value-order filter can be expressed as SELECT * FROM Orders WHERE paid_amount > 100. section 8.4.1
Listing 8.2 Selecting high value orders
SELECT * FROM Orders WHERE paid_amount > 100;
Shows a concise filtering query over an Orders relation. section 8.4.1
