Examkin

Question

Difficulty: HardCost-Optimized Database Design and Capacity Planning

A company is designing a cost-optimized database architecture for a non-critical internal application. The architecture includes two databases:

1. An Amazon RDS for MySQL database that has a predictable baseline workload of `500500` writes/second and `1,0001,000` reads/second. During a weekly `44-hour` batch processing job, the read workload spikes to `10,00010,000` reads/second. A short period of downtime is acceptable during database updates or failures.
2. An Amazon DynamoDB table that stores temporary session state data. The write traffic is highly volatile and unpredictable, spiking from `1010` writes/second to `8,0008,000` writes/second within seconds, followed by long periods of idle time.

Which two strategies should the solutions architect implement to meet these requirements in the most cost-effective manner? (Select TWO.)

  1. Configure the RDS for MySQL database as a single-Availability Zone (Single-AZ) instance, purchase a Reserved Instance to cover the baseline workload, and configure RDS Read Replicas with Application Auto Scaling to handle the weekly read spikes.Answer
  2. Configure the DynamoDB table to use On-Demand capacity mode.Answer
  3. C
    Configure the DynamoDB table with Provisioned Capacity Mode and enable Amazon DynamoDB auto-scaling to scale the write capacity between `1010` and `8,0008,000` write capacity units.
  4. D
    Configure the RDS for MySQL database as a Multi-Availability Zone (Multi-AZ) DB instance deployment, and direct the weekly batch processing read traffic to the standby DB instance.
  5. E
    Purchase a Compute Savings Plan to cover the baseline compute charges for both the RDS for MySQL database instance and the DynamoDB table.

Answer

To design a cost-optimized database tier, configure the RDS for MySQL database as a Single-AZ instance with a Reserved Instance contract for the baseline workload and configure RDS Read Replicas with Application Auto Scaling to handle the weekly spikes. Additionally, configure the DynamoDB table to use On-Demand capacity mode to accommodate the unpredictable write bursts.
For the relational database, a Single-AZ RDS for MySQL instance with a Reserved Instance contract is the most cost-effective way to handle the predictable baseline workload since high availability is not required and short downtime is acceptable. During the weekly batch read spikes, RDS Read Replicas with Application Auto Scaling can scale out to handle the read load and scale in when not in use. For the non-relational database, DynamoDB On-Demand capacity mode is the most cost-effective configuration because the workload is highly volatile and unpredictable, and On-Demand capacity adapts instantly to spikes without over-provisioning or throttling.

Step-by-Step Solution

1
Analyze the availability and deployment model requirements for the relational database.
Since the application is non-critical and can tolerate short downtime, a Single-AZ deployment is selected. This reduces database licensing and infrastructure costs by approximately `50%50\%` compared to a Multi-AZ deployment.
Avoids unnecessary high-availability standby costs when the business requirements do not demand it.
2
Select the purchasing option and scaling strategy for the predictable relational database workload.
A Reserved Instance contract is chosen for the stable baseline workload (`500500` writes/second and `1,0001,000` reads/second). For the weekly `44-hour` read spikes, RDS Read Replicas with Application Auto Scaling are configured.
Reserved Instances minimize baseline costs, while read replicas scale out dynamically only when needed, keeping idle costs at zero.
3
Determine the capacity planning mode for the volatile non-relational database workload.
DynamoDB On-Demand capacity mode is chosen for the session state table.
Since write spikes are sudden and unpredictable, auto-scaling would fail to react quickly enough, causing write failures, while provisioning for the peak of `8,0008,000` writes/second would lead to extreme idle capacity costs.

Key Concept

Cost optimization for databases involves matching deployment structures and capacity planning modes to workload predictability and availability requirements. Predictable baselines benefit from Reserved Instances, while temporary spikes should use auto-scaling or on-demand modes. Non-critical applications should avoid Multi-AZ deployments unless high availability is mandatory.
Estimated Time:2m 30s
Rate this question

Topics

Cost-Optimized Database Design and Capacity Planning