You currently have a single on-premises Kafka cluster in a data center in the us-east region that is responsible for ingesting messages from IoT devices globally. Because large parts of globe have…

Question

Options

• AEdge TPUs as sensor devices for storing and transmitting the messages.
• BCloud Dataflow connected to the Kafka cluster to scale the processing of incoming messages.
• CAn IoT gateway connected to Cloud Pub/Sub, with Cloud Dataflow to read and process the messages from Cloud Pub/Sub.
• DA Kafka cluster virtualized on Compute Engine in us-east with Cloud Load Balancing to connect to the devices around the world.

• A
  5% (2)
• B
  18% (7)
• C
  66% (25)
• D
  11% (4)

Explanation

Option C is correct because it replaces the single on-premises Kafka cluster with a fully managed, globally scalable stack: an IoT gateway (Cloud IoT Core or a regional MQTT broker) ingests device messages into Cloud Pub/Sub, which is designed to absorb massive, bursty traffic without manual scaling, and Cloud Dataflow provides a managed Apache Beam pipeline to process those messages reliably.

• A is wrong because Edge TPUs are machine-learning inference chips, not message-queuing or connectivity solutions - they don't solve the ingestion spike problem.
• B is wrong because adding Dataflow downstream of the existing Kafka cluster only helps with processing; it does nothing to absorb the upstream ingestion spikes hitting your on-prem cluster, and it keeps the expensive single-cluster bottleneck in place.
• D is wrong because simply virtualizing Kafka on Compute Engine and adding a load balancer moves the problem to the cloud but doesn't eliminate it - you still own cluster management, capacity planning, and cost spikes; it's lift-and-shift, not cloud-native.

Memory tip: Think "ingest → buffer → process" as the cloud-native IoT pattern. Pub/Sub is Google's infinitely scalable buffer (the "shock absorber" for bursty IoT traffic), and Dataflow is the managed processor - whenever a question mentions spiky, global IoT ingestion, that pairing is the Google-recommended answer.

No community discussion yet for this question.