--- name: azure-waf-performance-efficiency metadata: category: WellArchitectedFramework description: >- Generates performance efficiency guidance for Azure workloads based on the Performance Efficiency pillar of the Azure Well-Architected Framework. Use when setting performance targets, planning capacity, selecting service tiers, or designing scaling, partitioning and performance testing on Azure. --- # Azure Well-Architected Framework skill for the Performance Efficiency pillar ## Overview The Performance Efficiency pillar keeps every interaction efficient as demand grows. Every system has a limit beyond which it needs redesign or human work, so targets, capacity and scale units are decided in the design and tested in a production-like environment. ## Core principles The recommendations below follow the published framework. Written for this agent from Microsoft's published design review checklist. Codes are Microsoft's. - **Define performance targets** (PE:01): Set numeric targets, tied to requirements, for all flows. Grounding document: https://learn.microsoft.com/azure/well-architected/performance-efficiency/performance-targets - **Plan capacity** (PE:02): Plan ahead of seasonal change, releases, campaigns and events. Grounding document: https://learn.microsoft.com/azure/well-architected/performance-efficiency/capacity-planning - **Select the right services** (PE:03): Choose services and tiers that reach the targets and absorb expected change. Grounding document: https://learn.microsoft.com/azure/well-architected/performance-efficiency/select-services - **Measure consistently** (PE:04): Compare behaviour over time against baselines to find degradation. Grounding document: https://learn.microsoft.com/azure/well-architected/performance-efficiency/monitoring - **Optimize scaling and partitioning** (PE:05): Base both on the workload's scale unit. Grounding document: https://learn.microsoft.com/azure/well-architected/performance-efficiency/scale-partition - **Test performance** (PE:06): Test regularly in a production-like environment. Grounding document: https://learn.microsoft.com/azure/well-architected/performance-efficiency/performance-test - **Optimize code and infrastructure** (PE:07): Offload responsibilities to the platform. Grounding document: https://learn.microsoft.com/azure/well-architected/performance-efficiency/optimize-code-infrastructure - **Optimize data** (PE:08): Tune stores, partitions and indexes for actual use. Grounding document: https://learn.microsoft.com/azure/well-architected/performance-efficiency/optimize-data-performance - **Prioritize critical flows** (PE:09): Give the most important flows the most resources and attention. Grounding document: https://learn.microsoft.com/azure/well-architected/performance-efficiency/prioritize-critical-flows - **Optimize operational tasks** (PE:10): Limit the effect of scans, rotations, backups, reindexing and deployments. Grounding document: https://learn.microsoft.com/azure/well-architected/performance-efficiency/optimize-operational-tasks - **Respond to live issues** (PE:11): Set communication and responsibility, and learn from each event. Grounding document: https://learn.microsoft.com/azure/well-architected/performance-efficiency/respond-live-performance-issues - **Keep optimizing** (PE:12): Watch components that deteriorate over time, such as databases and networking. Grounding document: https://learn.microsoft.com/azure/well-architected/performance-efficiency/continuous-performance-optimize ## Relevant products The following are _examples_ of products relevant to this pillar: - **Scaling**: Autoscale, Virtual Machine Scale Sets, AKS cluster autoscaler, KEDA - **Delivery**: Azure Front Door, Azure CDN - **Data**: Azure Managed Redis, Cosmos DB partitioning, Azure SQL read replicas - **Testing**: Azure Load Testing, Application Insights ## Workload assessment questions Ask appropriate questions to understand the requirements and constraints of the workload and the organization. Choose from this list: - What are the numeric targets for latency, throughput and concurrency? - What is the expected peak, and when does it happen? - What is the workload's scale unit? - Which flows matter most under load? - Where is data read and written, and at what rate? - Where are users, relative to the workload's regions? - How is performance tested before release? ## Validation checklist Use this checklist to evaluate the architecture against this pillar: - Each critical flow has a numeric target. - Capacity is planned against a stated peak, and quotas are checked. - Service tiers are justified against the targets. - Scaling is automatic, with limits set. - Data partitioning and caching match the access pattern. - Performance baselines are recorded and compared. - Load tests run in a production-like environment. - Routine operations are scheduled away from peak.