OBSERVABILITY

Observability & monitoring

Metrics, logs, and events on one queryable SQL timeline — ingest, compress, alert, and expire in a single engine.

THE STATUS QUO

Telemetry stacks sprawl: a collector, a time-series database, a log store, and a dashboard layer — each with its own query language, retention config, and bill.

HOW EzraDB DOES IT

Land metrics and logs straight from your bus with exactly-once DEDUP KEY, backpressure, and dead-lettering — no collector-to-warehouse hop.

Compression designed for timestamps, gauges, and counters — telemetry shrinks to a fraction of raw size and still scans vectorized.

Raw telemetry expires automatically on the schedule you declare — no cron jobs deleting rows at 3am.

Service-level rollups and error-rate views stay current in under a second — your dashboards read a view, not a full scan.

tsquery + GIN indexes mean you grep your logs in SQL, in the same engine that holds your metrics.

Keep the last days in RAM, the last months on NVMe, and years in object storage — one table, one query surface.

*In development — not yet production-ready.

IN PRACTICE

One timeline, one query language, one system to page you — instead of four.

telemetry in, continuously
CREATE STREAMING PIPELINE tick_pipeline
  SOURCE KAFKA (brokers='localhost:9092', topic='ticks', group_id='ezradb')
  INTO si_ticks FORMAT JSON DEDUP KEY (sym, ts);