DOP-C02 Practice Question 87: Lambda Provisioned Concurrency Autoscaling
AWS Certified DevOps Engineer – Professional (DOP-C02) · Test 9 · question 7 of 10
A synchronous API backed by an AWS Lambda function suffers noticeable cold-start latency during sudden traffic spikes that do not follow a predictable daily schedule. The team enables provisioned concurrency on a published function alias to eliminate cold starts, but does not want to pay for a fixed, always-on amount of provisioned concurrency sized for peak load. They need the provisioned concurrency level to rise and fall automatically based on how much of it is actually being used at any given time. What should they configure?
Show answer and explanation
Answer:
Lambda is a registered scalable target in Application Auto Scaling, and a target tracking policy on LambdaProvisionedConcurrencyUtilization automatically raises provisioned concurrency when utilization runs high and lowers it when utilization drops, without a fixed always-on allocation and without predicting traffic in advance, which fits an unpredictable spike pattern.
A is wrong because it requires building custom alarm and remediation logic that duplicates a capability Application Auto Scaling already provides natively, and throttling is a lagging signal compared to a utilization-based target tracking policy. C is wrong because reserved concurrency only caps or guarantees a pool of regular concurrency, it does nothing to eliminate the cold starts that provisioned concurrency specifically addresses. D is wrong because EC2 Auto Scaling groups manage EC2 instances, they have no relationship to Lambda's provisioned concurrency, which is scaled through Application Auto Scaling's own Lambda integration.
Reference: Configuring provisioned concurrency for a function