You work for an online travel agency that also sells advertising placements on its website to other companies. You have been asked to predict the most relevant web banner that a user should see…

Question

Options

• AEmbed the client on the website, and then deploy the model on AI Platform Prediction.
• BEmbed the client on the website, deploy the gateway on App Engine, and then deploy the model
• CEmbed the client on the website, deploy the gateway on App Engine, deploy the database on
• DEmbed the client on the website, deploy the gateway on App Engine, deploy the database on

• A
  5% (2)
• B
  7% (3)
• C
  74% (32)
• D
  14% (6)

Explanation

Option C is correct because the three-tier architecture - client → App Engine gateway → database → AI Platform Prediction - satisfies all stated constraints simultaneously. The App Engine gateway is critical for security, as it prevents direct public exposure of the model endpoint while managing authentication and request routing. A database layer (typically Cloud Bigtable) is required to serve precomputed navigation-context features at the speed needed to meet the 300ms@p99 latency target across thousands of banner candidates.

Why the distractors fail:

• A skips the gateway entirely, exposing the model endpoint directly to the web client - a clear security violation.
• B adds the gateway (fixing the security issue) but omits the database, making it impossible to efficiently retrieve navigation-context features at scale within the latency budget.
• D likely differs from C in a detail that breaks either the latency, security, or simplicity requirement (e.g., a more complex serving infrastructure than necessary).

Memory tip: Think "Security → Speed → Simplicity." The gateway enforces security, the database enables speed (pre-fetched features beat real-time computation), and AI Platform Prediction is the managed, simplest-to-operate serving layer. If any tier is missing, one of these three pillars collapses.

No community discussion yet for this question.