cloudnative wiki

3.3-performance-improvement

1 min read

3.3 Performance Improvement

Optimization Tools

Compute Optimizer

  • Recommendations:
    • Over-provisioned: Downsize to save money.
    • Under-provisioned: Upsize to improve performance.
    • Optimized: Good balance.
  • Inference: Can predict EC2 instance types for ML inference.

CloudWatch for Performance

  • Custom Metrics: Push app-level metrics (e.g., “Page Load Time”, “Memory Used”) using the PutMetricData API or CloudWatch Agent.
  • Composite Alarms: Combine multiple alarms to reduce noise (e.g., “High CPU” AND “High IO”).

Acceleration & Edge

AWS Global Accelerator

  • Function: Directs traffic to optimal endpoints over the AWS global network.
  • IPs: Provided with 2 Static Anycast IPs.
  • Use Case: Non-HTTP traffic (TCP/UDP), gaming, IoT, or fast failover between regions (seconds vs minutes for DNS).

Edge Computing

  • CloudFront: Caching static content.
  • Lambda@Edge: Compute at Edge locations.
  • Outposts: Run AWS infrastructure on-premises.
  • Local Zones: Extensions of Regions closer to large population centers.
  • Wavelength: AWS infrastructure at 5G network edge.

WARNING

Exam Gotcha: Global Accelerator does not cache content. It accelerates the network path. CloudFront caches content.

Graph View