Term Kind Topic What it is
Tiered Storage Remote Storage Offload, Long-Term Log Retention, Hot-Cold Log Tiering pattern Streaming Cost Keeping recent log segments on broker local disk and older ones in object storage, which decouples retention from broker sizing and turns replay from an emergency measure into a routine design tool.
Transaction Log Stream concept CDC to Stream Change events derived from a database's write-ahead log — faithful to the table's mutations and to its internal model rather than to the business's.
Unbounded Dataset concept Streaming & Real-Time Data Input with no known end, which removes the option of waiting for completeness and forces every aggregate to be provisional.
Unbundled Event Store Disaggregated Telemetry Store, Object-Storage Column Store pattern Real-Time Analytical Stores A telemetry store that separates writing, compaction and querying into independently scaled services over shared object storage and a metadata store, so retention cost stops being tied to the size of the query…
Watermark Event-Time Progress, Completeness Estimate, Lateness Boundary concept Watermarks & Late Data A heuristic assertion that no further events earlier than a given event time are expected, which is how a stream processor decides when a window is complete - and therefore where the correctness-versus-promptn…
Watermark concept Watermarks & Late Data The stream processor's assertion that no further events older than a given event time are expected, which is what allows a window to close.
Watermark Lag metric Watermarks & Late Data How far behind the newest event a watermark is held, which trades result latency against how much late data is included.
Window Overlap Factor Hop Multiplier, Sliding Window Fan-Out metric Windowing The ratio of a sliding window's size to its advance - the number of windows each record belongs to, and therefore the multiplier on state, output volume and any naive sum taken downstream.
Windowing Strategy concept Windowing The choice of how a continuous stream is divided into finite groups for aggregation, which determines both the meaning of the result and the state it requires.