Pricing
Usage-based, and predictable before the invoice.
You pay for what you send and keep. The controls that decide how much that is are part of the product rather than an afterthought.
What you are billed on
Ingest
per GiB acceptedMeasured on what is accepted after policy, so data refused by a cardinality budget or an attribute deny list is never billed.
Retention
per GiB-dayStorage is rent rather than a purchase. Retention is set per environment, so staging need not be kept as long as production.
Active series
per 1k seriesThe cost a bytes meter cannot see: cardinality consumes index and memory rather than volume.
Query
per GiB scannedCosted before execution, so an expensive query is a decision rather than a surprise.
Estimate your volume
Rates are not published yet. This works out the quantities a rate would apply to, which is the part you cannot look up.
- Ingest, per month
- 1,500 GiB
- Held at steady state
- 1,500 GiB
Questions
- Why usage-based rather than per host?
- Per-host pricing charges the same for a quiet service and a noisy one, and it stops reflecting cost as soon as workloads are packed densely. Metering what actually costs us money — bytes accepted, bytes retained, series held — is the only model where the bill is predictable from what you send.
- What stops a bad deploy producing a large bill?
- Cardinality budgets and attribute policy are enforced as data arrives, not reconciled afterwards. An over-budget series is refused by name and surfaced in the portal, so the first place you hear about a runaway label is the product rather than the invoice.
- Are queries billed?
- Query usage is metered, and every query is costed before it runs. An expensive query is refused with its estimate rather than executed and charged for.
- Can I keep my existing tooling?
- Yes. ClueLake ingests OpenTelemetry and speaks PromQL for the supported query set, so existing dashboards and alert rules can keep working while telemetry lands here.