Examkin

Question

Difficulty: MediumCost-Optimized Database Design and Capacity Planning

A logistics company is designing the database tier for a new tracking application. The workload consists of two distinct tables:

1. A metadata store that receives a predictable, steady volume of updates throughout the day, averaging 500500 writes per second and rarely exceeding 600600 writes per second.
2. A tracking portal that experiences sudden, highly unpredictable spikes in read traffic during promotional events, scaling rapidly from 1010 reads per second to over 15,00015,000 reads per second.

To design a cost-optimized solution that guarantees performance, which two capacity planning configurations should a solutions architect recommend? (Select TWO.)

  1. Configure the metadata store table to use provisioned capacity mode with Auto Scaling enabled.Answer
  2. Configure the tracking portal table to use on-demand capacity mode.Answer
  3. C
    Configure the tracking portal table to use provisioned capacity mode with a fixed limit of 15,00015,000 read capacity units (RCUs).
  4. D
    Configure the metadata store table to use on-demand capacity mode.
  5. E
    Purchase a Compute Savings Plan to cover the read and write capacity unit costs of both DynamoDB tables.

Answer

Configure the metadata store table to use provisioned capacity mode with Auto Scaling enabled, and configure the tracking portal table to use on-demand capacity mode.
For the metadata store, the steady and predictable write rate of 500500 to 600600 writes per second is most cost-effectively handled using provisioned capacity with Auto Scaling, which charges a lower rate per unit of capacity. For the tracking portal, the sudden and unpredictable spikes up to 15,00015,000 reads per second are best handled by on-demand capacity mode. This prevents throttling during bursts and avoids paying for idle capacity during normal hours, resulting in a lower total cost.

Step-by-Step Solution

1
Analyze the workload characteristics of the metadata store.
The metadata store has a predictable, steady baseline of 500500 to 600600 writes per second.
Identifying the traffic patterns helps choose between provisioned and on-demand capacity modes.
2
Determine the cost-optimal capacity mode for the metadata store.
Provisioned capacity mode with Auto Scaling is chosen.
For a steady-state workload, provisioned capacity is significantly cheaper per request than on-demand capacity.
3
Analyze the workload characteristics of the tracking portal.
The tracking portal has extremely spiky and unpredictable read traffic, surging from 1010 to 15,00015,000 reads per second.
Sudden surges can cause throttling if scaling is not fast enough, or high costs if capacity is over-provisioned.
4
Determine the cost-optimal capacity mode for the tracking portal.
On-demand capacity mode is chosen.
On-demand capacity handles sudden, rapid spikes instantly without throttling and scales down to zero cost when idle.
5
Evaluate purchasing options and discounts.
Compute Savings Plans are ruled out for DynamoDB.
Compute Savings Plans do not apply to DynamoDB capacity costs, so other discount mechanisms (like DynamoDB Reserved Capacity) must be considered for the provisioned table if applicable.

Key Concept

Selecting between DynamoDB Provisioned Capacity (with Auto Scaling) and On-Demand Capacity based on workload predictability and spikiness to optimize costs.
Estimated Time:1m 30s
Rate this question

Topics

Cost-Optimized Database Design and Capacity Planning