You are working on an app that will save the device metrics produced every minute by various IoT devices. You decide to collect the data for two entities; the devices and the device metrics produced…

Question

Options

• ACreate a document with the deviceid property and other device data, and add a property called
• BCreate a document with the deviceid property and other device data. After that embed each
• CCreate a document with the deviceid property and other device data, and add a property called
• DNone of these.

• A
  3% (1)
• B
  9% (3)
• C
  76% (26)
• D
  12% (4)

Explanation

The correct answer is C: Create a document with the deviceid property and other device data, and add a property called 'type' (or similar discriminator) to distinguish between entity types within the same container. This is a standard Cosmos DB design pattern called 'multi-entity containers' or 'single-table design.' Since Cosmos DB is schema-agnostic, different entity types can coexist in the same container. A discriminator property (e.g., 'type': 'device' or 'type': 'metric') allows queries to filter by entity type. This approach reduces the number of containers, simplifies partition key design, and can lower provisioned throughput costs. Option B (embedding all metrics into the device document) would cause unbounded document growth as metrics accumulate every minute per device - this is an anti-pattern for high-frequency time-series data. Option A may suggest a similar approach but without the type discriminator, making querying harder.

No community discussion yet for this question.