Examkin

Question

Difficulty: HardAWS Pricing Models

A ride-sharing platform is planning its Amazon EC2 deployment strategy for three distinct workloads:

1. Core dispatch database servers that must run continuously with a predictable, steady-state workload profile for a multi-year period.
2. A fleet of application servers that scale dynamically to handle highly volatile, unpredictable spikes in passenger demand during rush hours, where active customer connections cannot be abruptly dropped.
3. A background data-cleansing pipeline that processes non-urgent passenger feedback surveys in batches, which is designed to tolerate interruptions and can resume paused tasks.

Which combination of Amazon EC2 pricing models is the most cost-effective and architecturally appropriate for these workloads?

  1. Reserved Instances for the core database servers, On-Demand Instances for the application servers, and Spot Instances for the data-cleansing pipelineAnswer
  2. B
    Spot Instances for the core database servers, Reserved Instances for the application servers, and On-Demand Instances for the data-cleansing pipeline
  3. C
    On-Demand Instances for the core database servers, Spot Instances for the application servers, and Reserved Instances for the data-cleansing pipeline
  4. D
    Reserved Instances for the core database servers, Spot Instances for the application servers, and On-Demand Instances for the data-cleansing pipeline

Answer

Reserved Instances for the core database servers, On-Demand Instances for the application servers, and Spot Instances for the data-cleansing pipeline
The combination of Reserved Instances for the core database servers, On-Demand Instances for the application servers, and Spot Instances for the data-cleansing pipeline is correct. The database servers require continuous, predictable coverage over multiple years, which is the exact use case for Reserved Instances to maximize discount. The application servers require uninterrupted runtimes during volatile spikes, making On-Demand Instances the appropriate choice to prevent dropped client connections. The data-cleansing pipeline is fault-tolerant and batch-oriented, allowing it to take advantage of the steep discounts offered by Spot Instances despite potential interruptions.

Step-by-Step Solution

1
Analyze the workload requirements for the database servers.
The core database servers run continuously with a predictable, steady-state profile for a multi-year period. This steady-state profile aligns with Reserved Instances (or Savings Plans), which offer significant discounts in exchange for a 1-year or 3-year commitment.
Matching steady-state, long-term workloads with commitment-based pricing models minimizes the baseline cost.
2
Analyze the workload requirements for the application servers.
The application servers experience highly volatile, unpredictable spikes during rush hours, and active customer connections cannot be dropped. Spot Instances are ruled out due to potential interruptions. Reserved Instances are ruled out because the demand is highly variable and temporary. Therefore, On-Demand Instances are required to ensure continuous availability during spikes.
On-Demand Instances provide reliable, non-interruptible capacity for unpredictable, short-term scaling needs.
3
Analyze the workload requirements for the data-cleansing pipeline.
The data-cleansing pipeline runs background batch jobs that are fault-tolerant and can tolerate interruptions. This makes it an ideal fit for Spot Instances, which offer spare AWS capacity at discounts of up to 90% compared to On-Demand pricing.
Leveraging Spot Instances for interruptible, non-urgent workloads provides the maximum cost savings without risking application stability.

Key Concept

Selecting optimal Amazon EC2 pricing models (On-Demand, Reserved Instances, Spot Instances) based on workload predictability, stability, commitment length, and fault tolerance.
Rate this question

Topics

AWS Pricing Models

More questions from this topic