DP-420 Practice Test

Sample DP-420 Practice Questions

10 real questions from the free DP-420bank. Pick an answer in your head, then open “Show answer” for the explanation.

1. 1. In Azure Cosmos DB for NoSQL, what does setting a container's DefaultTimeToLive property to -1 mean?

   • A. TTL is disabled on the container entirely
   • B. TTL is enabled on the container, but items do not expire by default unless an item sets its own ttl value
   • C. Items in the container expire after -1 seconds, treated as immediate expiration
   • D. The container will delete itself after -1 days
   Show answer

   Correct answer: B

   B is correct -- a DefaultTimeToLive of -1 turns TTL on with no default expiration, so items live forever unless they set their own item-level ttl. A describes a null/missing DefaultTimeToLive instead. C misinterprets -1 as a literal negative duration. D is not a real behavior.

2. 2. A tenant container in a pooled fleetspace has one physical partition with 7,000 RU/s of dedicated throughput, and a traffic spike needs as much capacity as possible. The fleetspace pool has 300,000 RU/s available and default limits apply. What is the maximum total RU/s that this physical partition can consume during the spike?

   • A. 7,000 RU/s, because pool RU/s become available once the container autoscale maximum is raised
   • B. 10,000 RU/s, because dedicated plus pool consumption is capped per physical partition
   • C. 12,000 RU/s, because each physical partition can add 5,000 RU/s from the pool to its dedicated RU/s
   • D. 300,000 RU/s, because pool RU/s are shared at the fleetspace level and are not bound to a partition
   Show answer

   Correct answer: B

   By default a physical partition can draw up to 5,000 extra RU/s from the pool, but its total dedicated plus pool consumption cannot exceed 10,000 RU/s, so the maximum is min(5,000 + 7,000, 10,000) = 10,000 RU/s. Pool RU/s are used automatically once dedicated RU/s are exceeded, without raising the autoscale maximum. Adding 5,000 to 7,000 ignores the 10,000 RU/s per-partition cap. Pool capacity is still subject to the standard physical partition limits, so the full pool cannot flow to one partition.

3. 3. A SaaS application partitions its Users container by CompanyID and wants to guarantee that no two users within the same company share an email address, while still allowing the same email address to be reused across different companies. Which unique key design achieves this?

   • A. Set /CompanyID as the unique key path
   • B. Set /email as the partition key instead of /CompanyID
   • C. Keep /CompanyID as the partition key and add /email as a unique key path, so uniqueness is enforced per logical partition
   • D. Create a separate container per company and set /email as that container's partition key
   Show answer

   Correct answer: C

   C is correct -- because unique keys are scoped to a logical partition, keeping CompanyID as the partition key and adding email as the unique key path enforces 'unique per company' exactly as required, while allowing the same email in a different company's partition. A would make CompanyID itself unique, which breaks having multiple users per company. B changes the partition strategy entirely and doesn't scope uniqueness per company. D is an unnecessarily complex architecture that isn't required to satisfy the stated requirement.

4. 4. Before choosing a partition key and deciding whether to embed or reference related data in Azure Cosmos DB for NoSQL, why is it important to first identify the application's data access patterns?

   • A. Because Cosmos DB requires access patterns to be registered before a container can be created
   • B. Because containers with documented access patterns support the change feed
   • C. Because the partition key and modeling choices should optimize for how the data will actually be read and written, not just how it is logically related
   • D. Because access patterns determine the container's maximum RU/s automatically
   Show answer

   Correct answer: C

   C is correct -- Microsoft's guidance is to design partitioning and embedding/referencing decisions around actual query and write patterns, since a technically valid model can still perform poorly if it doesn't match how the application reads and writes data. A, B, and D describe requirements or behaviors that don't exist in Cosmos DB.

5. 5. A stock-trading application models a person's portfolio by referencing stock items rather than embedding their full details, since stock prices change frequently throughout the day. What is the main tradeoff of this referenced design compared to embedding?

   • A. It makes the portfolio document larger as trades accumulate
   • B. It requires the container to use hierarchical partition keys
   • C. Reading a complete portfolio with current stock details requires multiple database requests instead of one
   • D. It prevents the application from ever updating the stock item
   Show answer

   Correct answer: C

   C is correct -- referencing trades read simplicity for write efficiency: displaying a portfolio now requires resolving each stock reference with an extra request, but a stock's price only needs to be updated in one place. A describes an embedding-related concern, not referencing. B is not a requirement of referencing. D is false -- referenced items can still be updated normally.