A declarative pipeline describes the tables you want and the queries that produce them. The platform uses those references to build a dependency graph. A streaming table incrementally processes input according to its streaming semantics. A materialized view stores a query result and refreshes it to reflect its sources. A regular view stores a query definition and reads its sources when queried.
Aster Supply needs a cleaned silver feed and a regional gold summary. Name each output explicitly so the graph shows which one depends on the other. In this browser, refreshes process a finite source snapshot and expose row counts. You can practice dependencies and data quality, but a local refresh cannot prove the latency or capacity of a production stream. Choose the output type from the required update semantics, not from which command name sounds faster.