AZ-305 Question 66
Single answerYou are designing an Azure SQL Database solution for a multi-tenant SaaS application that experiences unpredictable spikes in CPU utilization for short intervals, while usage is minimal during off-peak hours. Cost containment during low usage is a priority, but you also need the database to scale quickly when demand increases. Which service tier and compute tier combination should you recommend?
- A
Premium tier with provisioned compute
- B
Business Critical tier with serverless compute
- C
General Purpose tier with serverless compute
- D
Basic tier with provisioned compute
Show answer and explanation
Correct answer: C
Explanation
In Azure SQL Database, the serverless compute tier (available under the General Purpose service tier) automatically scales compute resources within configurable limits based on workload demand. This approach reduces costs during periods of low activity while enabling the database to handle sudden spikes in usage. For more details on Azure SQL Database service tiers and compute tiers, refer to the official Microsoft documentation: https://learn.microsoft.com/azure/azure-sql/database/service-tiers.
- A. Incorrect.
Premium tier with provisioned compute can handle high performance requirements but does not offer the auto-scaling serverless capability needed for unpredictable workloads. Costs remain higher even during off-peak times.
- B. Incorrect.
Business Critical tier does not currently support the serverless compute model. It focuses on low-latency, high-availability scenarios but would not provide the automatic scaling and cost benefits you require.
- C. Correct.
General Purpose tier with serverless compute is correct. The serverless option automatically scales resources based on demand, reducing costs during off-peak periods while handling short bursts of CPU-intensive activity.
- D. Incorrect.
Basic tier with provisioned compute offers a cost-saving entry point but is limited in performance and does not scale automatically. It’s best for small, less critical workloads without significant performance spikes.