MLA-C01 Practice Question 78: Endpoint Multi AZ Availability
AWS Certified Machine Learning Engineer – Associate (MLA-C01) · Test 8 · question 8 of 10
A lender's credit-scoring model runs on a SageMaker AI real-time endpoint with a single ml.c5.2xlarge instance. Last month an instance failure made the endpoint unavailable for about 40 minutes while it recovered, and loan applications stopped. The team wants the endpoint to keep serving through the failure of a single instance or Availability Zone, with the simplest change.
What should the team do?
Show answer and explanation
Answer:
AWS recommends deploying multiple instances for every production endpoint. When an endpoint has more than one instance, SageMaker AI automatically tries to distribute them across Availability Zones, so an instance failure or AZ outage leaves the remaining instances serving traffic. If the endpoint runs in a VPC, its subnets should span at least two AZs. Changing the instance count is a small update to the existing endpoint (option C).
- A is wrong because a second Region with DNS failover protects against a Regional outage, which is far more than this requirement needs. It doubles the infrastructure and is the opposite of the simplest change. - B is wrong because one larger instance is still one instance. When it fails, the endpoint is down exactly as before. - D is wrong because recreating an endpoint after it fails is *recovery*, not availability. The outage lasts at least as long as provisioning takes.
Reference: Best practices for deploying models on SageMaker AI Hosting Services