Временные ряды с RedisTimeSeries

Storing metrics (CPU, temperature, request counts) as plain keys wastes memory and makes range queries awkward. RedisTimeSeries is a module purpose-built for timestamped numeric data, with compression, retention, and downsampling.

TS.CREATE makes a new series. You can set retention (how long to keep samples) and labels for filtering.

TS.ADD appends a timestamped value. Use * for the current server time, or supply an explicit millisecond timestamp.

If a series does not exist, TS.ADD can create it on the fly, optionally with retention and labels, which is handy for high-cardinality ingest.

TS.RANGE returns samples between two timestamps. Use - and + for the minimum and maximum bounds.

Add AGGREGATION to bucket and summarize samples, for example averaging into 60-second windows.

• Functions: avg, min, max, sum, count

A compaction rule automatically writes aggregated values from a source series into a lower-resolution destination, keeping long-term data cheap.

TS.MRANGE queries multiple series at once, filtered by labels, ideal for dashboards showing all sensors in a room.

The RETENTION setting (in milliseconds) auto-expires old samples. Combine short retention on raw series with long retention on downsampled series for a tiered storage strategy.

TS.INFO reports sample count, memory, retention, labels, and any compaction rules attached to a series.

RedisTimeSeries shines for monitoring, IoT telemetry, and real-time analytics where you need fast appends, range queries, and automatic downsampling without a separate time-series database.

You used RedisTimeSeries to create series with retention and labels, add samples, query ranges with aggregation, downsample via compaction rules, query many series with TS.MRANGE, and inspect them with TS.INFO. It is a compact way to handle metrics and telemetry inside Redis.