Amazon EC2 instances over-provisioned for Microsoft SQL Server

This Recommendation identifies EC2 instances running Microsoft SQL Server that have more vCPUs than needed for SQL Server licensing, presenting opportunities to rightsize to smaller instance types and reduce costs.

What it does

This Recommendation analyzes your EC2 instances running Microsoft SQL Server and identifies cases where:

Instances have more vCPUs than the maximum recommended for the SQL Server workload
The instance can be rightsized to a smaller, less expensive instance type
SQL Server licensing costs can be reduced by using instances with fewer vCPUs

AWS Trusted Advisor provides:

Current instance type and vCPU count
Maximum recommended vCPU count based on workload analysis
Recommended instance type to rightsize to
Estimated monthly savings from rightsizing

Why it matters

Cost Savings: Rightsizing over-provisioned instances can lead to significant infrastructure savings
License Optimization: SQL Server licensing is typically core-based; fewer vCPUs mean lower licensing costs
Efficiency: Running appropriately sized instances improves resource utilization across your infrastructure
Performance: Properly sized instances can actually improve performance by reducing unnecessary resource overhead

Recommended actions

Review Recommendations: Examine each flagged instance, noting:
- Current instance type and vCPU count
- Maximum recommended vCPU count
- Recommended instance type
- Estimated monthly savings
- SQL Server edition in use
Analyze Workload Patterns: Before rightsizing:
- Review CPU utilization metrics over time
- Identify peak usage periods
- Verify that the recommended instance size can handle peak loads
- Consider seasonal or cyclical workload variations
Plan the Migration: Develop a rightsizing strategy:
- Prioritize instances with highest savings potential
- Group similar workloads for batch processing
- Schedule changes during maintenance windows
- Prepare rollback procedures
Test in Non-Production: Before production changes:
- Test the recommended instance type in dev/staging environments
- Validate SQL Server performance with realistic workloads
- Verify application behavior under load
- Test backup and recovery procedures
Execute Rightsizing: During scheduled maintenance:
- Stop the EC2 instance
- Change the instance type to the recommended size
- Start the instance and verify SQL Server starts correctly
- Test application connectivity
- Monitor performance for anomalies
Monitor Post-Change: After rightsizing:
- Track CPU and memory utilization closely
- Monitor SQL Server performance metrics
- Verify application response times remain acceptable
- Document any issues and adjustments made

Additional Considerations

Burst Workloads: If your workload has significant burst patterns, ensure the recommended instance type can handle peak loads
Licensing Compliance: Verify that downsizing maintains compliance with Microsoft SQL Server licensing requirements
High Availability: Consider the impact on your HA/DR strategy when changing instance types
Stop/Start Impact: Changing instance types requires stopping the instance, which will cause downtime
Elastic IPs: If using Elastic IPs, they will be retained when changing instance types
Instance Store: If using instance store volumes, be aware that data will be lost when stopping the instance (EBS-backed volumes are preserved)
Performance Testing: Always conduct thorough performance testing before and after rightsizing to ensure the new instance type meets your requirements