Question 38
Domain 1: Plan and Manage an Azure AI SolutionAn Azure AI Language resource in East US is experiencing intermittent high latency (3+ seconds) during business hours. The application requires sub-500ms response times. What should you investigate first, and what is the likely fix?
Correct answer: B
Explanation
Azure AI Language resources are subject to throughput limits, so intermittent spikes during business hours often indicate the service is throttling at its transactions-per-second limit. The first thing to check is whether the resource has hit its TPS cap; if so, the fix is to scale up the tier or distribute traffic with regional load balancing to keep latency under 500 ms.
Why each option is right or wrong
A. Enable HTTPS — the API calls are being transmitted unencrypted
B. Check if the resource has hit its transactions-per-second (TPS) limit; scale up the tier or add regional load balancing
Azure AI Language resources are governed by SKU-based throughput caps, so intermittent 3+ second delays during peak business hours are a classic sign of throttling once the transactions-per-second ceiling is reached. The first check should be the resource’s TPS/throughput metrics and any 429/throttling signals; if the cap is the issue, the remedy is to move to a higher tier with more capacity or spread requests across multiple regions to reduce per-resource load and keep latency below 500 ms.
C. Enable Managed Identity to reduce authentication overhead
D. Compress the input documents before sending them