On-demand vs provisioned cost comparison in DynamoDB - Quick Revision & Key Differences

Recall & Review

beginner

What is the main difference between On-demand and Provisioned capacity modes in DynamoDB?

On-demand capacity charges you based on the actual read and write requests you make, while provisioned capacity requires you to specify the number of reads and writes per second you expect, paying for that capacity whether you use it or not.

beginner

How does On-demand capacity mode help with unpredictable workloads?

On-demand capacity automatically scales to handle sudden increases or decreases in traffic, so you only pay for what you use without needing to manage capacity settings.

intermediate

What is a potential cost benefit of using Provisioned capacity mode?

If your workload is steady and predictable, provisioned capacity can be cheaper because you pay a fixed rate for reserved capacity, which can be less expensive than paying per request in On-demand mode.

intermediate

What happens if your application exceeds the provisioned capacity in DynamoDB?

Requests that exceed the provisioned capacity may be throttled, causing delays or errors unless you have enabled auto scaling or increased capacity.

beginner

Which capacity mode is better for a new app with unknown traffic patterns?

On-demand capacity mode is better because it adjusts automatically to traffic changes and you only pay for what you use, avoiding the risk of over or under provisioning.

Which DynamoDB capacity mode charges based on actual read/write requests?

A Reserved capacity

B Provisioned capacity

C Burst capacity

D On-demand capacity

On-demand capacity charges you only for the read and write requests you actually make.

What is a risk of using provisioned capacity without auto scaling?

A Requests being throttled when traffic spikes

B Paying too much for unused capacity

C Automatic scaling to meet demand

D No cost for unused capacity

If traffic exceeds provisioned capacity, requests can be throttled causing errors.

When is provisioned capacity usually more cost-effective?

A For unpredictable workloads

B For very low traffic apps

C For steady, predictable workloads

D For apps with no traffic

Provisioned capacity is cheaper when you have steady traffic and can reserve capacity accordingly.

Which capacity mode requires you to specify read and write units ahead of time?

A On-demand

B Provisioned

C Auto scaling

D Burst

Provisioned capacity requires you to set the number of read and write units before use.

What is a benefit of On-demand capacity mode?

A No need to manage capacity settings

B Fixed monthly cost

C Cheaper for steady workloads

D Requires manual scaling

On-demand mode automatically adjusts capacity without manual management.

Explain the cost differences between On-demand and Provisioned capacity modes in DynamoDB.

Think about how you pay for electricity: per use or fixed monthly fee.

Describe scenarios when you would choose On-demand capacity over Provisioned capacity in DynamoDB.

Consider when you don’t want to guess how much capacity you need.

Practice

1. Which DynamoDB billing mode charges you only for the read and write requests you actually use, without any upfront capacity planning?

easy

A. Provisioned mode

B. Reserved mode

C. On-demand mode

D. Manual mode

Solution

1. Step 1: Understand billing modes in DynamoDB

   DynamoDB offers two main billing modes: on-demand and provisioned. On-demand charges per request without pre-allocating capacity.

2. Step 2: Identify the mode with no upfront capacity planning

   On-demand mode does not require capacity planning and bills you based on actual usage.

3. Final Answer:

   On-demand mode -> Option C

4. Quick Check:

   On-demand charges per request = On-demand mode [OK]

Hint: On-demand means pay per request, no setup needed [OK]

Common Mistakes:

• Confusing provisioned mode as pay-per-request
• Thinking reserved mode exists in DynamoDB billing
• Assuming manual mode is a billing option

2. Which of the following is the correct way to specify provisioned capacity in a DynamoDB table creation command?

easy

A. ProvisionedThroughput: { ReadCapacityUnits: 5, WriteCapacityUnits: 5 }

B. BillingMode: 'ON_DEMAND'

C. CapacityUnits: { Read: 5, Write: 5 }

D. Throughput: { Read: 5, Write: 5 }

Solution

1. Step 1: Recall DynamoDB table creation syntax for provisioned mode

   Provisioned mode requires specifying ProvisionedThroughput with ReadCapacityUnits and WriteCapacityUnits.

2. Step 2: Match the correct syntax

   ProvisionedThroughput: { ReadCapacityUnits: 5, WriteCapacityUnits: 5 } uses the exact correct keys: ProvisionedThroughput with ReadCapacityUnits and WriteCapacityUnits.

3. Final Answer:

   ProvisionedThroughput: { ReadCapacityUnits: 5, WriteCapacityUnits: 5 } -> Option A

4. Quick Check:

   Provisioned capacity uses ProvisionedThroughput [OK]

Hint: Provisioned capacity uses ProvisionedThroughput keys [OK]

Common Mistakes:

• Using BillingMode for provisioned capacity
• Incorrect key names like CapacityUnits or Throughput
• Confusing ON_DEMAND with provisioned syntax

3. Consider a DynamoDB table with provisioned capacity of 10 read units and 5 write units. If your application performs 8 strongly consistent reads and 3 writes per second, which billing mode will likely cost less?

medium

A. Cannot determine without data size

B. On-demand mode

C. Both cost the same

D. Provisioned mode

Solution

1. Step 1: Analyze usage vs provisioned capacity

   The app uses 8 reads and 3 writes per second, which is below the provisioned 10 reads and 5 writes capacity.

2. Step 2: Compare cost implications

   Provisioned mode charges fixed capacity regardless of usage, so if usage is steady and below provisioned, it is cheaper than on-demand which charges per request.

3. Final Answer:

   Provisioned mode -> Option D

4. Quick Check:

   Steady low usage < provisioned = Provisioned cheaper [OK]

Hint: If usage steady and below capacity, provisioned is cheaper [OK]

Common Mistakes:

• Assuming on-demand is always cheaper
• Ignoring provisioned capacity limits
• Thinking cost depends only on data size

4. You set a DynamoDB table to provisioned mode with 5 read and 5 write capacity units, but your app suddenly needs 20 reads per second. What is the likely problem and how to fix it?

medium

A. Provisioned capacity too low; increase capacity or switch to on-demand

B. On-demand mode is enabled; switch to provisioned mode

C. Table is in on-demand mode; reduce read requests

D. Provisioned capacity too high; reduce capacity units

Solution

1. Step 1: Identify the mismatch between capacity and usage

   The app needs 20 reads per second but provisioned capacity is only 5 reads, causing throttling.

2. Step 2: Determine the fix

   To fix, increase provisioned capacity or switch to on-demand mode which scales automatically.

3. Final Answer:

   Provisioned capacity too low; increase capacity or switch to on-demand -> Option A

4. Quick Check:

   Low provisioned capacity causes throttling; increase or switch [OK]

Hint: If usage > provisioned, increase capacity or use on-demand [OK]

Common Mistakes:

• Confusing on-demand and provisioned modes
• Reducing capacity when usage is high
• Ignoring throttling symptoms

5. A startup expects unpredictable traffic spikes but wants to control costs. They plan to use DynamoDB. Which billing strategy best balances cost and flexibility?

hard

A. Use provisioned mode with high capacity to handle spikes

B. Use provisioned mode with low capacity and enable auto scaling

C. Use on-demand mode but limit requests manually

D. Use on-demand mode to pay only for actual requests

Solution

1. Step 1: Understand startup needs

   They want to handle unpredictable spikes but also control costs.

2. Step 2: Evaluate billing options

   Provisioned mode with auto scaling adjusts capacity automatically, balancing cost and flexibility better than fixed provisioned or pure on-demand.

3. Step 3: Compare with other options

   High fixed provisioned is costly; on-demand can be expensive at scale; manual limits reduce flexibility.

4. Final Answer:

   Use provisioned mode with low capacity and enable auto scaling -> Option B

5. Quick Check:

   Auto scaling balances cost and spikes well [OK]

Hint: Auto scaling provisioned mode balances cost and spikes [OK]

Common Mistakes:

• Choosing high fixed provisioned capacity unnecessarily
• Assuming on-demand always cheaper for spikes
• Trying to limit requests manually in on-demand mode