MSFTREL04-BP01 Use Amazon EC2 Auto Scaling in combination with Application Auto Scaling
Microsoft workloads often experience variable demand patterns that require dynamic resource scaling to maintain performance and cost efficiency. Applications like SharePoint farms, SQL Server databases, and .NET web applications face fluctuating user loads, batch processing requirements, and seasonal traffic variations that make static resource provisioning ineffective and costly.
Desired outcome: Implement automated scaling mechanisms that combine instance-level and application-level auto scaling capabilities to optimize resource utilization. Configure predictive scaling using historical data and machine learning to anticipate capacity needs, while maintaining regulatory requirements through dedicated host allocation. Automate capacity management through system tooling that responds to monitoring metrics and scheduled events.
Common anti-patterns:
-
Relying on human intervention to adjust resources in response to demand changes, leading to slow reactions and potential over- or under-provisioning.
-
Maintaining a fixed resource capacity based on peak demand expectations, resulting in wasted resources during low-demand periods and potential shortages during unexpected spikes.
-
Focusing solely on infrastructure scaling without considering the need for application-level scaling, potentially leading to bottlenecks in database connections, caching, or other application components.
Benefits of establishing this best practice:
-
Dynamically adjusts resource capacity to actual demand, reducing waste from over-provisioning while reducing performance issues from under-provisioning.
-
Automatically responds to changes in workload patterns and system failures, maintaining service availability without manual intervention.
-
Uses historical data and machine learning to anticipate and prepare for demand spikes before they occur, reducing reactive scaling delays.
-
Maintains licensing and regulatory requirements through controlled scaling of dedicated resources while still achieving operational efficiency.
-
Reduces administrative overhead through automated capacity management, freeing up teams to focus on higher-value activities while providing consistent performance.
Level of risk exposed if this best practice is not established: High
Implementation guidance
Configure auto scaling groups to dynamically adjust compute resources based on demand patterns. Enable predictive scaling using historical metrics to proactively manage capacity before peak loads occur. For applications with specific licensing requirements, use dedicated hosts with auto scaling policies to maintain adherence. Implement automation tools to handle routine capacity management tasks, using monitoring alerts and scheduled actions to control resource scaling.
Begin by establishing baseline metrics and performance thresholds for both infrastructure and application components. Configure scaling policies that combine predictive and dynamic adjustments, using historical data to anticipate capacity needs while maintaining real-time responsiveness to unexpected changes. Set up monitoring and alerting to track scaling activities, and implement gradual scaling steps to avoid resource thrashing. Document scaling decisions and regularly review performance data to refine thresholds and policies, providing optimal resource utilization while maintaining service levels and regulatory requirements.
Implementation steps
-
Configure EC2 Auto Scaling groups with appropriate launch templates and scaling policies for Windows instances, including SQL Server and IIS workloads.
-
Enable predictive scaling using AWS Auto Scaling with machine learning forecasts, incorporating Microsoft workload patterns and licensing considerations for dedicated hosts.
-
Set up Amazon CloudWatch Application Insights to monitor .NET and SQL Server applications, automatically detecting anomalies and providing application-level metrics for scaling decisions across IIS, SQL Server, and application servers.
-
Create CloudWatch alarms and custom metrics for Microsoft applications including SQL Server connection counts, IIS request queues, and .NET application performance counters to trigger scaling actions, using Application Insights automated dashboards.
-
Implement application-level scaling automation for SQL Server Always On Availability Groups, IIS worker process scaling, and SharePoint farm expansion using Systems Manager automation documents triggered by Application Insights events.
-
Configure session state management and connection pooling strategies to support horizontal scaling of .NET applications and SQL Server workloads across multiple instances, using Application Insights correlation data for optimization.
Resources
Related documents:
Related tools:
