Microsoft Certified: Azure AI Engineer AssociatePlan and manage an Azure AI solutionMedium
A company is developing an Azure AI solution that uses Azure Cognitive Search. They anticipate varying workloads throughout the day, with peak hours requiring significantly more search queries per second. To optimize costs while maintaining performance, they want to ensure the search service can automatically scale out during peak demand and scale in during off-peak hours. Which scaling feature should be configured for the Azure Cognitive Search service?
- AEnable autoscaling for replicas.
- BUpgrade the pricing tier to a higher capacity.
- CProvision more replicas manually.
- DIncrease the partition count.
Show answer & explanationAnswer & explanation
Correct answer: A. Enable autoscaling for replicas.
Azure Cognitive Search supports autoscaling for replicas, allowing the service to automatically adjust the number of replicas based on demand. This ensures optimal performance during peak loads and cost savings during off-peak times by scaling in.
Why the other options are wrong
- B. Upgrading the pricing tier provides a fixed increase in capacity but doesn't offer dynamic scaling in and out based on varying demand, leading to potential over-provisioning during off-peak hours.
- C. Provisioning replicas manually requires constant monitoring and intervention, which is not suitable for 'automatically scale out... and scale in'.
- D. Increasing the partition count scales for document storage and indexing throughput, not primarily for query throughput, and it's a manual adjustment.
Azure Cognitive Search Scaling
Azure Cognitive Search scales using partitions (for storage and indexing) and replicas (for query throughput). Replicas can be autoscaled to handle varying query loads.
- Partitions scale storage and indexing.
- Replicas scale query throughput.
- Autoscaling for replicas adjusts capacity based on demand.
- Manual scaling or tier upgrades are less dynamic and cost-efficient for variable loads.
Memory trick: Replicas handle your 'requests', and 'Auto' makes them flexible.