Read and write capacity units in DynamoDB - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ query_result

intermediate

Calculating Read Capacity Units for Strongly Consistent Reads

You have a DynamoDB table with items of size 3 KB each. You want to perform 100 strongly consistent read operations per second. How many read capacity units (RCUs) do you need to provision?

A 100 RCUs

B 200 RCUs

C 300 RCUs

D 400 RCUs

💡 Hint

Remember that one strongly consistent read capacity unit can read up to 4 KB per second.

Each strongly consistent read capacity unit (RCU) can read up to 4 KB per second. Since each item is 3 KB, one RCU can handle one read per second. For 100 reads per second, you need 100 RCUs. Because the item size is less than 4 KB, each read consumes 1 RCU. So total RCUs = 100 * (3 KB / 4 KB rounded up) = 100 * 1 = 100. The question expects the calculation for strongly consistent reads, so the correct answer is 100 RCUs.

❓ query_result

intermediate

Write Capacity Units for Large Items

You want to write 50 items per second to a DynamoDB table. Each item is 2 KB in size. How many write capacity units (WCUs) do you need to provision?

A 150 WCUs

B 50 WCUs

C 100 WCUs

D 200 WCUs

💡 Hint

One write capacity unit can write up to 1 KB per second.

Each write capacity unit (WCU) can write 1 KB per second. Since each item is 2 KB, each write consumes ceil(2/1) = 2 WCUs. For 50 writes per second, total WCUs = 50 * 2 = 100.

📝 Syntax

advanced

Identify the Error in Capacity Unit Calculation Expression

Which of the following expressions correctly calculates the number of read capacity units needed for eventually consistent reads of 150 items per second, each 5 KB in size?

A RCUs = 150 * CEIL(5 / 4) / 2

B RCUs = 150 * CEIL(5 / 4) * 2

C RCUs = 150 / 2 * CEIL(5 / 4)

D RCUs = CEIL(150 * 5 / 4) / 2

💡 Hint

Eventually consistent reads consume half the RCUs of strongly consistent reads.

For eventually consistent reads, the formula is: RCUs = (number_of_reads * ceil(item_size/4 KB)) / 2. Option A correctly applies this formula. Option A multiplies by 2 instead of dividing. Option A changes the order incorrectly. Option A divides after ceiling, which is incorrect.

🔧 Debug

advanced

Debugging Provisioned Capacity Calculation for Mixed Reads

A developer wrote this formula to calculate total RCUs for 100 strongly consistent reads and 200 eventually consistent reads per second, each item 3 KB:

total_RCU = (100 + 200) * CEIL(3 / 4) / 2

What is the problem with this formula?

DynamoDB

total_RCU = (100 + 200) * CEIL(3 / 4) / 2

A It should add 2 to the reads before multiplying

B It incorrectly divides the sum of reads by 2, applying eventually consistent factor to all reads

C It should use item size 4 KB instead of 3 KB

D It should multiply by 2 instead of dividing by 2

💡 Hint

Strongly consistent reads consume full RCUs, eventually consistent reads consume half.

The formula divides the total reads by 2, which incorrectly applies the eventually consistent read factor to both strongly and eventually consistent reads. The correct approach is to calculate RCUs separately for each type and then sum them.

🧠 Conceptual

expert

Optimizing Capacity Units for Variable Item Sizes

You have a DynamoDB table with items of varying sizes: 50% are 1 KB, 30% are 3 KB, and 20% are 7 KB. You expect 100 writes per second. How should you provision write capacity units (WCUs) to minimize cost while avoiding throttling?

A Provision WCUs based on median item size (3 KB) for all writes: 100 * 3 = 300 WCUs

B Calculate weighted average item size and provision WCUs accordingly: (0.5*1 + 0.3*3 + 0.2*7) * 100 = 3.2 * 100 = 320 WCUs

C Provision WCUs based on the smallest item size (1 KB) for all writes: 100 * 1 = 100 WCUs

D Provision WCUs based on the largest item size (7 KB) for all writes: 100 * 7 = 700 WCUs

💡 Hint

DynamoDB charges WCUs per 1 KB chunk per write. Underprovisioning causes throttling.

Because item sizes vary, and DynamoDB charges WCUs per 1 KB chunk, you must provision capacity for the largest item size to avoid throttling. Using averages or medians risks underprovisioning and throttling. So provisioning 700 WCUs (100 writes * 7 KB) is safest.