Observability
TrafficPeak
Zuplo's gateway layer sits between every client and your backend — which means it sees every request. TrafficPeak (powered by Akamai Hydrolix) gives you a queryable, real-time analytics store built on that data. Every request Zuplo processes is streamed as a structured log event, available for dashboards, alerts, and ad-hoc queries within seconds of arrival.
What Zuplo can tell you
Because the gateway processes every request, it has visibility that backend services don't — including rejected requests, rate-limited consumers, and edge-level geographic context.
Request volume & throughput
Total requests per second, peak traffic windows, and volume trends by endpoint or consumer. Know your traffic shape before you need to scale.
Latency breakdown
p50, p95, and p99 latency per route — split between gateway processing time and upstream response time. Pinpoint whether slowness is in your gateway policies or your backend.
Error rates and status codes
4xx and 5xx rates by endpoint, consumer, and region. Spike detection lets you catch a broken deploy or an abusive client before it escalates.
Per-consumer analytics
Attach consumer IDs via a TypeScript logging policy and get per-consumer usage breakdowns — who's calling what, at what volume, and with what error rate.
Geographic distribution
Traffic origin by city, country, and ASN. Identify underserved regions, latency outliers by geography, and anomalous patterns by network.
Policy outcomes
Log which requests were rate-limited, rejected by auth, or transformed. Audit the gateway's decision-making, not just the traffic it passed through.
What gets logged
Core fields are captured automatically. Custom fields are added via a TypeScript logging policy on any route — attach any value derivable from the request or response context.
Request metadata
Consumer context
Geographic & network
Custom (TypeScript)
Custom fields via TypeScript: Add a logging policy to any route and return any object. Consumer plan tier, upstream service name, extracted JWT claims — all land in the same log stream and are immediately queryable in TrafficPeak.
How TrafficPeak works
Zuplo batches log events and streams them asynchronously to Akamai TrafficPeak's Hydrolix-powered columnar store. The pipeline is off the request path — no latency is added to API responses.
Request processed
Zuplo handles the request normally. Logging runs asynchronously — zero latency impact.
Events batched and streamed
Log events are batched at configurable intervals and streamed to Hydrolix at high throughput.
Queryable within seconds
Events land in the columnar store and are immediately available for dashboards and ad-hoc queries.