Usage Metering
Usage metering without the pipeline.
Capture every billable event. Materialize running totals. Query on demand. No Kafka, no batch jobs, no stale data.
Without Primatomic
Kafka → consumer → aggregator → cache → billing API
Batch jobs that run hourly, data always slightly stale
Revenue leakage from dropped events or double-counting
Separate pipeline per billing metric
With Primatomic
Append event → WASM view → query
Continuous materialization, read-after-write
Immutable log, exactly-once fold processing
One log, multiple views for different metrics
How it works
Four calls to production usage metering
Create a usage log
One log per customer, billing entity, or whatever maps to your pricing model.
Append usage events
Emit events from your API gateway, app layer, or anywhere usage happens.
Deploy billing views
Upload WASM functions for monthly totals, current period usage, rate limit checks.
Query at billing time
Pull aggregated usage for invoicing, rate limiting, or real-time dashboards.
# Create a usage log
POST /logs
-d '{"name": "customer-acme-usage"}'
→ {"log_id": "550e8400..."}
# Append a usage event
POST /logs/$LOG_ID/append
-d '{"type": "api.call", "endpoint": "/v1/generate",
"tokens_in": 150, "tokens_out": 420,
"model": "gpt-4"}'
→ {"sequence": 1}
# Deploy a billing view
POST /logs/$LOG_ID/views/monthly-totals
--data-binary @monthly_totals.wasm
→ {"view_id": "b2c3d4e5..."}
# Query usage for invoicing
POST /logs/$LOG_ID/views/monthly-totals/query
?month=2025-03
→ {"gpt-4": {"tokens_in": 1482000, "tokens_out": 3291000}}Why Primatomic
No batch lag.
Views update continuously. Your billing dashboard and rate limiter always reflect the latest event.
Auditable by default.
Every billable event is in the immutable log. Billing disputes become "let me query the log" instead of "let me check the pipeline."
Multiple views, one log.
Monthly totals, per-minute rate checks, cost allocation by team — all derived from the same event stream without duplicating data.
How Primatomic compares
| Primatomic | Kafka + custom pipeline | Orb / Amberflo | |
|---|---|---|---|
| Integration complexity | One API call per event | Kafka, consumers, aggregators | SDK integration |
| Custom aggregation | Any WASM function | Your code, your infra | Their supported dimensions |
| Real-time accuracy | Read-after-write | Eventual, depends on lag | Near-real-time |
| Auditability | Immutable event log | If you built it | Vendor-managed |
| Raw event access | Always, forever | Depends on retention | Limited |
Meter usage without the pipeline.
Start on the free tier. No credit card required.