Microsoft Certified: Azure Cosmos DB Developer Specialty - (DP-420) Exam Questions

Correct Answer: B

In Azure Cosmos DB, the consistency levels range from strong to eventual, with several levels in between that offer various trade-offs between performance, availability, and consistency. When choosing the weakest consistency level possible for an application, you would typically want to select the level that meets the application's requirements while providing the highest performance and lowest latency.

Based on the requirements provided for each application in the scenario:

1. Reporting: Must be able to track the total order counts within five minutes of orders being placed.

   • This requirement indicates that some staleness is acceptable but should be within a bounded timeframe. The appropriate consistency level would be Bounded Staleness, which guarantees that reads are no more stale than a specified amount of time or number of versions (updates).

2. Purchasing: Must guarantee that the latest committed stock quantities are used always.

   • This requirement calls for the most recent data to be read, without any staleness. The consistency level that guarantees this is Strong, which ensures that the read data is the most recent committed version.

3. Fulfillment: Must ensure that orders are read in the order in which they are placed.

   • This requirement suggests that the order of operations must be preserved. The consistency level that provides a guarantee of the order of operations is Consistent Prefix, which ensures that reads never see out-of-order writes.

So, the consistency levels for each application should be set as follows:

• Reporting: Bounded Staleness

• Purchasing: Strong

• Fulfillment: Consistent Prefix

These choices satisfy the applications' requirements using the weakest consistency level possible, which can help maximize performance and availability.

Correct Answer: C

In Azure Cosmos DB, there are specific limits for both request units per second (RU/s) and storage, which are defined per partition:

1. Maximum RU/s per physical partition:

   • The maximum RU/s that can be provisioned for a single physical partition is 10,000 RU/s. This is the upper limit set by Azure Cosmos DB.

2. Maximum storage per logical partition:

   • Each logical partition has a maximum storage limit of 20 GB. This means that the data within a logical partition should not exceed this size to maintain optimal performance and to ensure that operations do not fail due to exceeding partition storage limits.
     

Therefore, the appropriate options for the partition limits for request units per second (RU/s) and storage are:

• Maximum RU/s per physical partition: 10,000

• Maximum storage per logical partition: 20 GB

Correct Answer: A


In Azure Cosmos DB, you can use the Time to Live (TTL) feature to have items automatically deleted after a certain time period. The TTL is set in seconds, so to have an item deleted after two hours, you would set the TTL to 7200 seconds (2 hours * 60 minutes * 60 seconds).

The item definition should include the "ttl" property set to 7200. If the container is set up to respect individual item TTLs, this will cause the item to be deleted after 7200 seconds.

Based on the options in the screenshot and the requirement, you should set:

• "ttl": 7200
  

This ensures that the item will expire after two hours. If the TTL on the container level is disabled, then it needs to be enabled for item-level TTL to work. If the container-level TTL is set to -1, it means that TTL is enabled but items will not expire unless they have their own TTL set. If it's set to a specific value, items without their own TTL will inherit that value, but can be overridden by setting an item-specific TTL.

Correct Answer: A

Answer 1: 5,000 RU/s.

• The total provisioned throughput of 30,000 RU/s must be divided among the physical partitions. If there are 6 partitions, each partition would support 30,000 RU/s6=5,000 RU/s630,000 RU/s​=5,000 RU/s. This is the correct choice because it correctly divides the total throughput by the number of partitions to determine the throughput capacity of each individual partition.

Answer 2: 60,000 RU/s.

• Azure Cosmos DB allows a container to scale up its throughput to a point before it triggers a partition split. The choice of 60,000 RU/s suggests that the container can handle up to double its initially provisioned throughput of 30,000 RU/s without needing to split partitions to maintain performance. This is the correct choice because it reflects the capacity of the container to scale its throughput to handle increased load, which is a common practice in distributed databases like Cosmos DB to ensure that performance scales with demand.

Correct Answer: B, D
When configuring a partition key for an Azure Cosmos DB container, the goal is to ensure that the partition key provides efficient distribution of data and requests across physical partitions to optimize performance and cost.

Here's how the characteristics listed would impact your partition key choice:

A. Unique

• A unique partition key would ensure that each item has its own partition, but this would not minimize the size of the partition; in fact, it would maximize it. Additionally, it could lead to inefficient use of RU/s and potential hotspots if the read requests are not evenly distributed.

B. High Cardinality

• A high cardinality means there are many unique values, which can help distribute the data and requests more evenly across partitions. This usually helps to prevent bottlenecks on a single partition and can minimize the size of each partition by spreading data out. It also reduces maintenance effort since the distribution is handled by the database based on the partition key's natural variance.

C. Low Cardinality

• A low cardinality means there are fewer unique values, which could lead to some partitions being much larger than others if the distribution of data is skewed. This can create hotspots and is not ideal for minimizing partition size or maintenance effort.

D. Static

• A static characteristic means the partition key does not change over time. This is crucial for minimizing maintenance because changing partition keys can be complex and resource-intensive. It also ensures that the partition key remains consistent, which is good for read-heavy workloads since it allows for efficient data retrieval.
  

Given your requirements to minimize the size of the partition and minimize maintenance effort, the best characteristics to prioritize would be:

B. High Cardinality

• For even distribution and efficient use of resources.

D. Static

• To ensure the partition key remains unchanged over time, which is important for reducing maintenance efforts.

So, the correct answers are B (high cardinality) and D (static).

Correct Answer: A

Updated-

To ensure that the read operations of App1 can request either bounded staleness or consistent prefix consistency in Azure Cosmos DB, you should modify the following options for each consistency level:

Bounded Staleness:

• The account level options

Consistent Prefix:

• The request level options

This configuration allows you to specify these consistency levels in the context of your application’s read operations while adjusting them at the appropriate levels in Cosmos DB.

The correct answers are:

• For "Bounded Staleness": The request level options.

• For "Consistent Prefix": The request level options.

Based on the consistency level requirements for App1, here's how to modify each consistency level:

For Bounded staleness:

• Select "The account level options"
  Since bounded staleness is a stronger consistency level than session (the account's default), it must be configured at the account level.

For Consistent prefix:

• Select "The request level options"
  Since consistent prefix is a weaker consistency level than session (the account's default), it can be configured at the request level.

This configuration follows Azure Cosmos DB's consistency level hierarchy where:

• Stronger consistency levels (Strong, Bounded Staleness) must be set at account level.

• Equal or weaker consistency levels (Consistent Prefix, Session, Eventual) can be set at request level.

• The default consistency level of session allows for request-level configuration of weaker consistency models.

The consistency levels in Azure Cosmos DB from strongest to weakest are:

1. Strong

2. Bounded Staleness

3. Session (default)

4. Consistent Prefix

5. Eventual

The key rule is that you can only relax consistency (move to weaker levels) at the request level.

To use stronger consistency levels like Bounded Staleness, you must configure it at the account level.

Correct Answer: C

Based on the consistency requirements for each app, here's how you should match the consistency levels:

• App1: Requires that reads must always return the most recent committed version of an item, where the commit occurred during the same session. This requirement matches the description of Session consistency. Session consistency ensures that a client always reads its own writes, which is what App1's requirement specifies.

• App2: Needs to always return the most recent committed version of an item, even if the commit occurred during another session. This requirement fits Strong consistency, which guarantees that reads will always reflect the most recent write across all sessions.

• App3: Aims to minimize write latency and can tolerate data staleness. Eventual consistency would be the appropriate choice here as it provides the lowest latency and allows for some staleness in the data, which aligns with the app's requirements.
  

Here is the recommended default consistency level for each app:

• App1: Session consistency

• App2: Strong consistency

• App3: Eventual consistency
  

These recommendations are based on the typical behavior of Azure Cosmos DB consistency levels, ensuring that each app's requirements are met while minimizing concurrency issues.

Correct Answer: B, D
The Azure Cosmos DB SQL SDK identifies the optimal endpoint for reading and writing based on several factors.

Two of the key factors that can influence the SDK's endpoint selection are:

1. B. Regional Latency: The SDK is designed to choose the optimal endpoint with the least amount of latency. This ensures that read and write operations are as fast as possible, which is crucial for performance in distributed applications.

   
   

2. D. The PreferredLocations configuration: This configuration allows you to specify a list of regions, ordered by preference, that the SDK should target for read and write operations. By setting PreferredLocations, you can control the regional failover order, thereby directing traffic to the nearest or most appropriate region based on your application's needs.

   
   

The consistency level set at the account or through RequestOptions (A and C) does not directly influence the SDK's choice of the optimal endpoint; rather, it dictates the consistency behavior of read and write operations. The availability of a region (E) could affect the endpoint selection if the preferred region is not available, causing the SDK to failover to the next available region as specified by the PreferredLocations configuration or the default failover order.

These insights are derived from understanding how the Azure Cosmos DB SDKs function in terms of connectivity modes, error handling, and client configuration best practices, as explained in the Microsoft Azure Cosmos DB documentation. The SDKs are designed to be resilient and automatically manage connections to provide the best possible performance and reliability for your applications.

Correct Answer: C
DefaultTimeToLive to null

This is because setting DefaultTimeToLive to null on a container in Azure Cosmos DB means that the items in the container will never expire. This disables the TTL policy at the container level, so no automatic deletion of items based on time-to-live will occur. If you were to set a numeric value, it would specify the number of seconds after which items would expire, and setting it to -1 would mean that items could have individual TTL values but by default would not expire. However, to ensure that no items ever expire regardless of individual item settings, you would set the container's DefaultTimeToLive to null

Correct Answer: D

For the given scenario, where writes from IoT devices occur every second and the data is retained indefinitely, the partition key selection is critical for performance and scalability. Here's a review of the options:

• Option A (deviceManufacturer): This is not ideal because it's likely that many devices are from the same manufacturer, which could lead to hot partitions due to many writes going to the same partition.

• Option B (synthetic key with deviceId and timestamp): While this would distribute writes across partitions and minimize hot spots due to the high granularity, it could lead to an extremely high number of partitions, each with little data, which can be inefficient and may not be the best use of Cosmos DB resources.

• Option C (synthetic key with deviceId and deviceManufacturer): This is also not ideal as it combines two low-cardinality fields and doesn't solve the problem of potential hot partitions, especially if the manufacturer has a large number of devices.

• Option D (deviceId): Using deviceId as the partition key is typically a good choice because it has high cardinality, ensuring that writes are distributed across many partitions. Each device would write to its own partition, which avoids hot partitions. However, if some devices send data much more frequently than others, this could still lead to hot partitions.

Given these considerations, Option D (deviceId) is generally the best choice because it is likely to have the highest cardinality and will distribute the writes across partitions effectively. Each IoT device will have its own partition, which minimizes partition skew and avoids the capacity limits associated with having too many writes directed at a single partition. However, careful monitoring is advised to ensure that no single device generates enough writes to cause a hot partition