Can anyone clarify the rate limiting thresholds for the predictive routing metrics API when aggregating data across multiple BYOC trunks?
We are running a daily analytics job that pulls queue.queued and queue.answered metrics for 15 trunks distributed across Asia/Singapore and Asia/Tokyo regions. The job executes via the Python SDK v2.14.0 using a service account with full analytics scopes. Starting last Tuesday, we hit 429 Too Many Requests errors consistently after querying roughly eight trunks. The retry logic in our script handles the backoff, but it pushes the job completion time from 45 minutes to over two hours, which breaks our SLA for morning reporting.
The documentation states:
“Rate limits for analytics endpoints are applied per tenant and are not aggregated across regions. Standard plans allow 1000 requests per minute.”
This seems contradictory given the behavior. The requests are sequential, not parallel, and each request targets a different queueId associated with a specific trunk. We have verified that the OAuth token is valid and has not expired. Is there a hidden throttle on cross-region queries, or are we hitting a per-endpoint limit that isn’t clearly documented? We need to understand if this is a platform constraint or a misconfiguration in our request structure.