Skip to main content

AWS Cost Estimation 2025 - Gaming Platform RGS

10K Concurrent Requests Architecture

Executive Summary

This document provides a comprehensive AWS cost estimation for the Remote Gaming System (RGS) platform capable of handling 10,000+ concurrent spin requests. The architecture follows AWS best practices with multi-AZ deployment for high availability and scalability.


Architecture Overview

Infrastructure Components

  • 1 VPC - Virtual Private Cloud with multi-AZ deployment
  • 1 Application Load Balancer - Auto-failover and health checks
  • 1 Internet Gateway - Internet connectivity (free)
  • 2 Public Subnets - One per availability zone
  • 2 Private Subnets - One per availability zone (for security)
  • 2 NAT Gateways - High availability internet access for private subnets
  • 1 ECS Fargate Cluster - Serverless container orchestration
  • 3 ECS Services - Game Backend, RGS, CMS Backend
  • 1 MemoryDB Redis Cluster - High-performance caching
  • 1 MongoDB Atlas Cluster - Dedicated database cluster
  • 3 ECR Repositories - Container image storage

Detailed Cost Breakdown

1. Compute Costs (ECS Fargate)

Task Configuration for 10K Concurrent Requests

  • Game Backend Service: 6-8 tasks
  • RGS Service: 4-6 tasks
  • CMS Service: 2-3 tasks
  • Total Tasks: 12-17 tasks running simultaneously

Resource Allocation per Task

  • vCPU: 2 vCPU per task
  • Memory: 4 GB per task
  • Platform: Linux/x86

Fargate Pricing (US East - N. Virginia)

  • vCPU Cost: $0.04048 per vCPU per hour
  • Memory Cost: $0.004446 per GB per hour

Monthly Compute Costs

Conservative Estimate (12 tasks):
- vCPU: 12 tasks × 2 vCPU × $0.04048 × 730 hours = $710.41
- Memory: 12 tasks × 4 GB × $0.004446 × 730 hours = $155.78
- Total: $866.19/month

High Load Estimate (17 tasks):
- vCPU: 17 tasks × 2 vCPU × $0.04048 × 730 hours = $1,006.33
- Memory: 17 tasks × 4 GB × $0.004446 × 730 hours = $220.68
- Total: $1,227.01/month

Average Monthly Compute Cost: $1,046.60

2. Load Balancer Costs

Application Load Balancer (ALB)

  • Hourly Charge: $0.0252 per hour
  • LCU Usage: ~2 LCUs for 10K concurrent requests
  • LCU Rate: $0.008 per LCU-hour

Monthly ALB Costs

- Hourly Charges: $0.0252 × 730 hours = $18.40
- LCU Charges: 2 LCUs × $0.008 × 730 hours = $11.68
- Total ALB Cost: $30.08/month

3. Networking Costs

NAT Gateway Costs (2 Gateways for HA)

  • Hourly Rate: $0.045 per NAT Gateway per hour
  • Data Processing: $0.045 per GB processed
  • Estimated Data: 2 TB/month through NAT Gateways

Monthly Networking Costs

- NAT Gateway Hours: 2 × $0.045 × 730 hours = $65.70
- Data Processing: 2,000 GB × $0.045 = $90.00
- Cross-AZ Data Transfer: 500 GB × $0.02 = $10.00
- Total Networking Cost: $165.70/month

4. Database & Cache Costs

MongoDB Atlas (AWS Dedicated Cluster)

  • Cluster Size: M40 (Dedicated)
  • Configuration: Multi-region with backup
  • Estimated Cost: $500-600/month

MemoryDB Redis Cluster

  • Node Type: r6g.large (2 vCPU, 12.93 GiB)
  • Configuration: 2 nodes for HA
  • Estimated Cost: $180-220/month

Monthly Database Costs

- MongoDB Atlas M40: $550.00
- MemoryDB Redis: $200.00
- Total Database Cost: $750.00/month

5. Container Registry (ECR)

Storage Requirements

  • Game Backend Image: ~1.5 GB
  • RGS Image: ~1.2 GB
  • CMS Backend Image: ~1.3 GB
  • Total Storage: ~4 GB
  • Version History: ~20 GB total

Monthly ECR Costs

- Storage: 20 GB × $0.10 = $2.00/month
- Data Transfer: Included within region
- Total ECR Cost: $2.00/month

6. Data Transfer Costs

Internet Data Transfer

  • Outbound to Internet: ~1 TB/month
  • Rate: $0.09/GB for first 10 TB

Monthly Data Transfer Costs

- Internet Transfer: 1,000 GB × $0.09 = $90.00/month

Total Monthly Cost Summary

ComponentMonthly Cost
ECS Fargate Compute$1,046.60
Application Load Balancer$30.08
Networking (NAT + Cross-AZ)$165.70
MongoDB Atlas$550.00
MemoryDB Redis$200.00
ECR Storage$2.00
Internet Data Transfer$90.00
Total Monthly Cost$2,084.38
Total Annual Cost$25,012.56

Scaling Scenarios

Low Traffic (5K Concurrent Requests)

  • Compute Tasks: 8-10 tasks
  • Monthly Cost: ~$1,650
  • Annual Cost: ~$19,800

Peak Traffic (15K Concurrent Requests)

  • Compute Tasks: 20-25 tasks
  • Monthly Cost: ~$2,800
  • Annual Cost: ~$33,600

Enterprise Scale (25K+ Concurrent Requests)

  • Compute Tasks: 35-40 tasks
  • Monthly Cost: ~$4,200
  • Annual Cost: ~$50,400

Cost Optimization Recommendations

1. Reserved Capacity (Up to 50% Savings)

  • Fargate Savings Plans: 1-3 year commitments
  • Potential Savings: $6,000-12,000 annually

2. Fargate Spot (Up to 70% Savings)

  • Use Cases: Non-critical workloads, batch processing
  • Potential Savings: $3,000-7,000 annually

3. VPC Endpoints (Reduce NAT Gateway Costs)

  • S3 VPC Endpoint: Free gateway type
  • DynamoDB VPC Endpoint: Free gateway type
  • Potential Savings: $1,200-2,400 annually

4. Right-Sizing and Auto-Scaling

  • CPU-based auto-scaling: Scale down during low traffic
  • Scheduled scaling: Predictable traffic patterns
  • Potential Savings: $2,000-4,000 annually

5. Database Optimization

  • MongoDB Atlas: Consider reserved instances
  • MemoryDB: Use data tiering for cost efficiency
  • Potential Savings: $1,800-3,600 annually

Regional Pricing Comparison (Annual Costs)

RegionTotal Annual CostDifference
US East (N. Virginia)$25,012Baseline
US West (Oregon)$25,512+$500
EU (Ireland)$27,014+$2,002
EU (Frankfurt)$28,215+$3,203
Asia Pacific (Singapore)$31,018+$6,006

Cost Monitoring & Alerts

  1. Budget Alerts: Set at $2,500/month (20% buffer)
  2. Cost Anomaly Detection: Automated unusual spend alerts
  3. Reserved Instance Recommendations: Monthly optimization reviews
  4. Resource Tagging: Track costs by service and environment

Key Metrics to Monitor

  • Cost per 1K requests: Target < $0.25
  • Monthly growth rate: Monitor for unexpected increases
  • Resource utilization: Keep CPU/Memory > 70%
  • Data transfer efficiency: Optimize cross-AZ communication

Conclusion

The estimated annual cost of $25,012 for handling 10K+ concurrent requests represents a competitive and scalable solution. With proper optimization strategies, costs can be reduced by 30-40%, bringing the annual expenditure down to $15,000-17,500.

This architecture provides enterprise-grade reliability, security, and performance while maintaining cost efficiency through AWS's pay-as-you-use model.