A gaming company is launching a new multiplayer game and needs to deploy two workloads on Amazon EC2. The first workload is a leaderboard analysis microservice that processes telemetry data in batches; this job is highly flexible, can run at any time of day, and is completely fault-tolerant to unexpected node terminations. The second workload is a matchmaking backend service that must run continuously to process player requests, requiring constant and predictable CPU performance over a one-year period. Which two pricing models should the gaming company select to minimize costs?
- Spot Instances for the leaderboard analysis microserviceAnswer
- Reserved Instances for the matchmaking backend serviceAnswer
- COn-Demand Instances for the matchmaking backend service
- DDedicated Hosts for the leaderboard analysis microservice
- ESpot Instances for the matchmaking backend service
Answer
Spot Instances for the leaderboard analysis microservice and Reserved Instances for the matchmaking backend service
For the leaderboard analysis microservice, the workload is flexible and fault-tolerant, making it a perfect fit for Spot Instances, which offer the highest discounts in exchange for potential interruptions. For the matchmaking backend service, the workload is predictable and runs continuously over a one-year period, making Reserved Instances the most economical choice due to the commitment discount they offer.
Step-by-Step Solution
Key Concept
Selecting optimal Amazon EC2 pricing models based on workload characteristics (interruptible vs. steady-state)