Polling vs webhook calculator
Requests per day and delay for polling, and what retries do to webhook delivery during an outage of your endpoint.
Model, not a measurement. Events arrive evenly. One outage of your endpoint per day. The retry preset is Aurora's documented approximate intervals, read as counted from the previous failed attempt. Check your vendor's own schedule.
| Polling | |
|---|---|
| Requests per day, one changed-since list | - |
| Requests per day, every project polled | - |
| Average delay / worst delay (minutes) | - |
| Share of list polls with nothing new (at least) | - |
| Webhook | |
| Requests per day, including retries | - |
| Events hit by the outage per day | - |
| Of those, delivered within 1 hour / later / not delivered | - |
| Average wait after your endpoint recovers (minutes) | - |
| Retry window (hours) | - |
After the same outage, a poller that resumes sees every change within one polling interval. That is why many teams add a slow reconcile poll to a webhook.