oriflux-sdk — API analytics for Python
pip install oriflux-sdk (PyPI, MIT license).
from oriflux_sdk import OrifluxMiddleware
app.add_middleware(
OrifluxMiddleware,
api_key="ofx_ing_…", # the API source's ingest key
endpoint="https://in.oriflux.sponge-theory.dev", # default
)
Guarantees
- Zero host impact: in-memory aggregation in 60 s windows (the
Apitally pattern), fire-and-forget delivery with a circuit breaker — an
Oriflux outage or a quota
429never surfaces in your app. - < 1 ms of overhead per request, bounded memory (~2,000 aggregation
keys per window, overflow into a
geo=unresolvedbucket). - Privacy: the caller’s IP is part of the aggregation key, resolved to country/ASN at ingestion then discarded — it is never persisted.
What you get
Volumes, latencies (p50/p95/p99), 4xx/5xx error rates, endpoints, consumers, caller geography — per minute.