Time as a First-Class Dimension

Time-series systems treat time as a primary design axis. Ingestion, storage layout, indexing, compression, and query execution are all built around the assumption that data arrives as a stream of observations ordered in time.

A time-series record carries the state of a process at a specific instant: CPU load on a node, vibration amplitude on a turbine, voltage on a feeder line, latency of a REST endpoint. New observations arrive as append-only writes, often at fixed sampling intervals or triggered by events. Historical data remains largely immutable, while queries scan contiguous time ranges: “last 5 minutes”, “previous 30 days”, “compare this week with last week”.…