API Usage
Settings → API Usage shows how much of your tenant's API capacity you're using — both the calls coming into PeakCommerce and the calls PeakCommerce makes out to your connected systems — measured against your own limits, so you can spot pressure before it becomes a problem.
One set of limits, both directions
PeakCommerce governs your tenant's own capacity — not your connected systems' limits. Every call is counted, whether it's inbound (a person or an API key calling PeakCommerce) or outbound (PeakCommerce calling Zuora, Salesforce, your identity provider, a fraud check, and so on). Inbound and outbound are tracked separately behind the scenes, but every number on this page is the combined total — inbound + outbound together. You never manage them as two separate budgets.
PeakCommerce doesn't mirror your providers' limits. It doesn't try to pre-guess Zuora's (or any system's) rate or concurrency ceilings. If a connected system is busy, it tells us — an HTTP
429 / Retry-After— and PeakCommerce backs off and retries, reacting to the real signal rather than a guessed one. The limits on this page are yours, set on your tenant.
Rate — UI, API, and AUTH
Traffic is grouped into three categories, each shown as requests per minute, hour, and day (RPM / RPH / RPD), and each measured against your tenant's limit:
API — programmatic / integration traffic. Combines inbound API-key calls and PeakCommerce's outbound calls to systems like Zuora and Salesforce.
AUTH — sign-in and identity traffic. Combines inbound authentication and PeakCommerce's outbound calls to your identity provider (Auth0, Okta). (AUTH is not inbound-only.)
UI — traffic from people using the portals in a browser.
Concurrency — Default, Query, and High-volume
Alongside rate, PeakCommerce limits how many calls run at the same time, in three buckets — again combining inbound and outbound, each shown against your tenant's limit:
High-volume — commerce and money-path work: placing orders, changing subscriptions, taking payments and refunds. (Outbound examples: creating a Zuora order or subscription, a payment, a fraud check.)
Query — large-result reads: lists, searches, and queries. (Outbound examples: Zuora ZOQL / Data Query, catalog list syncs, Salesforce SOQL.)
Default — single-record reads, metadata, sign-in, and everything else. (Outbound examples: fetching one Zuora account or subscription, an identity-provider token.)
Splitting concurrency into buckets keeps a burst of one kind of work — say a large catalog sync — from starving another, like a customer trying to check out.
Your limits are yours — and adjustable
These limits come from your tenant's governance tier and can be raised or tuned per tenant by your account team. If a legitimate workload is bumping a limit — a nightly sync, a burst of checkouts — that's a signal to right-size the limit, not a wall you can't move. (It's also why you shouldn't see "integration is over its concurrency limit" rejections: PeakCommerce no longer self-throttles against a guessed provider ceiling.)
Usage trends and per-integration detail
Beyond the limit gauges, the page shows usage over time and per connected system, so you can:
establish a baseline for normal traffic and treat departures from it as worth investigating;
read a sustained climb as capacity planning (or a runaway integration);
read a sudden drop as a possibly broken or disconnected integration.
For per-endpoint outbound detail — exact call counts, latency, and errors for each provider — pair this page with API Call Stats.
Gotchas
Every number is combined (inbound + outbound). A spike in API could be inbound API-key traffic, outbound Zuora/Salesforce calls, or both — use API Call Stats to see which.
AUTH includes outbound. Calls out to your identity provider (Auth0 / Okta) count as AUTH, not just inbound sign-ins.
A provider's own limit is separate. If Zuora or Stripe throttles one of our calls, that surfaces as their error in API Call Stats — it isn't governed by the numbers on this page.
History is retained for a bounded window, so very old periods may not be available.
Related
API Call Stats — per-endpoint counts, latency, errors, and per-provider outbound volume.
Last updated
Was this helpful?