Usage & limits

Unravel charges for history, projects, and seats - never for volume. You are not metered per event, and you never get a surprise bill for a busy day. What a plan does carry is a set of limits that keep the shared service healthy and predictable.

What counts as an event

An event is a single span: one HTTP request, one SQL query, one cache operation, one queued job, one outgoing HTTP call, one log line, one exception. A single user request to your app usually produces many events; a page with an N+1 problem alone can emit hundreds. The SDK batches events and ships them together, so an event is the unit of telemetry, not a unit of billing.

Retention

Telemetry is kept for your plan's window and then deleted:

Querying beyond your window returns nothing for the older range - the data is gone, not hidden.

Projects, seats, environments

Projects and seats scale with the plan; environments are always free: a project can have as many (production, staging, …) as you like at no extra cost. See the pricing page for the per-plan project and seat counts.

Fair-use ingest limits

To protect the shared service from runaway or abusive volume, ingestion is rate-limited per project, counted in events per minute:

These ceilings are deliberately generous. A healthy application sits orders of magnitude below them (our own dogfood app averages roughly a hundred events a minute). They are a technical safeguard against pathological spikes, not a billing meter, and the numbers are indicative guidance that may change.

If sustained traffic exceeds the ceiling, the SDK receives an HTTP 429 and drops the excess rather than queuing it - your application is never blocked or slowed. If you consistently approach the ceiling, get in touch: the higher tiers exist for genuinely high-volume apps, and Enterprise goes beyond the published ceilings (bring-your-own database is coming for it, too).