You need to design an architecture that serves asynchronous predictions to determine whether a particular mission-critical machine part will fail. Your system collects data from multiple sensors…

Question

Options

• A
  1. HTTP requests are sent by the sensors to your ML model, which is deployed as a microservice
• B
  1. Events are sent by the sensors to Pub/Sub, consumed in real time, and processed by a
• C
  1. Export your data to Cloud Storage using Dataflow.
• D
  1. Export the data to Cloud Storage using the BigQuery command-line tool

• A
  13% (4)
• B
  73% (22)
• C
  10% (3)
• D
  3% (1)

Explanation

The architecture needs to be asynchronous and handle streaming sensor data that must be aggregated over a 12-hour window before inference. Pub/Sub (B) is the right entry point: sensors publish events to Pub/Sub, decoupling data collection from processing. A Dataflow streaming pipeline consumes these events, maintains rolling 12-hour sensor averages, and triggers the ML model to predict failures in the next N minutes. Predictions can be written to a database for downstream consumption. Direct HTTP requests to a microservice (A) are synchronous, not asynchronous. Exporting to Cloud Storage via Dataflow (C) or the BigQuery CLI (D) are batch patterns that cannot support real-time, sensor-driven streaming with sub-minute latency for mission-critical failure prediction.

No community discussion yet for this question.