Deployment Strategies MACHINE-LEARNING-PROFESSIONAL Practice Questions (Page 3)

Questions 11–15

1. 11application · medium

   A fintech company needs to update its fraud-detection model serving endpoint. The ML team wants to minimize downtime and be able to revert to the previous model instantly if the new model's error rate spikes. They have enough compute resources to run both versions simultaneously. Which deployment strategy best meets these requirements?

   • ARecreate deployment
   • BShadow deployment
   • CBlue-green deployment
   • DCanary deployment
   Reveal Solution Discussion Select an answer first
2. 12application · medium

   A large e-commerce platform wants to deploy a new recommendation model to its high-traffic serving endpoint. The team wants to monitor the new model's performance on a small percentage of real traffic before rolling it out fully, and they need to be able to adjust the traffic split dynamically. Which deployment strategy should they use?

   • ARolling deployment
   • BCanary deployment
   • CBlue-green deployment
   • DRecreate deployment
   Reveal Solution Discussion Select an answer first
4. 13expert · hard

   A media streaming service must deploy a new content-ranking model. The team values both rapid rollback and the ability to validate the model on a subset of users. They have limited compute resources and cannot afford to run two full production environments simultaneously. Which deployment strategy best balances these constraints?

   • ARolling deployment
   • BBlue-green deployment
   • CCanary deployment
   • DShadow deployment
   Reveal Solution Discussion Select an answer first
5. 14application · medium

   A machine learning engineer is managing a Databricks Model Serving endpoint with two versions of a sentiment-analysis model. They want to route 10% of traffic to the new version and 90% to the current version. What is the correct way to configure this in Databricks Model Serving?

   • ADeploy the new version as the only version and rely on the model to handle all traffic
   • BCreate a second endpoint for the new version and use a load balancer to distribute 10% of traffic
   • CSet the traffic percentage for the new version to 10 and the current version to 90 in the endpoint's configuration
   • DUse a feature flag in the application code to randomly route 10% of requests to the new endpoint
   Reveal Solution Discussion Select an answer first
6. 15foundation · easy

   Why is canary deployment particularly suitable for high-traffic applications?

   • AIt requires only one environment, reducing infrastructure costs significantly.
   • BIt guarantees zero downtime because the old version is decommissioned before the new one is deployed.
   • CIt eliminates the need for monitoring because the traffic shift is automatic.
   • DIt allows the new version to be tested with a small, real-user subset, limiting the blast radius of potential issues.
   Reveal Solution Discussion Select an answer first