Auto Scaling Endpoints Quiz

5 questions Pass: 70% +25 pts (50 with Pro)

Quiz covering ML Infrastructure

What you'll be asked

  1. What is the primary purpose of implementing auto-scaling for machine learning model endpoints?
  2. Which metric is most commonly used as a trigger for scaling ML inference endpoints when latency is a critical business requirement?
  3. What is the main risk of setting a 'scale-in' cooldown period that is too short?
  4. When deploying a deep learning model that requires a GPU, why is 'target tracking' scaling often preferred over simple 'step scaling'?
  5. You observe that your auto-scaled endpoint frequently fails to handle sudden 'burst' traffic despite having a high maximum instance limit. What is the most likely architectural bottleneck?

Answers and explanations come after you submit.