Search for Well Architected Advice
-
Operational Excellence
-
- Resources have identified owners
- Processes and procedures have identified owners
- Operations activities have identified owners responsible for their performance
- Team members know what they are responsible for
- Mechanisms exist to identify responsibility and ownership
- Mechanisms exist to request additions, changes, and exceptions
- Responsibilities between teams are predefined or negotiated
-
- Executive Sponsorship
- Team members are empowered to take action when outcomes are at risk
- Escalation is encouraged
- Communications are timely, clear, and actionable
- Experimentation is encouraged
- Team members are encouraged to maintain and grow their skill sets
- Resource teams appropriately
- Diverse opinions are encouraged and sought within and across teams
-
- Use version control
- Test and validate changes
- Use configuration management systems
- Use build and deployment management systems
- Perform patch management
- Implement practices to improve code quality
- Share design standards
- Use multiple environments
- Make frequent, small, reversible changes
- Fully automate integration and deployment
-
Security
-
- Evaluate and implement new security services and features regularly
- Automate testing and validation of security controls in pipelines
- Identify and prioritize risks using a threat model
- Keep up-to-date with security recommendations
- Keep up-to-date with security threats
- Identify and validate control objectives
- Secure account root user and properties
- Separate workloads using accounts
-
- Analyze public and cross-account access
- Manage access based on life cycle
- Share resources securely with a third party
- Reduce permissions continuously
- Share resources securely within your organization
- Establish emergency access process
- Define permission guardrails for your organization
- Grant least privilege access
- Define access requirements
-
- Build a program that embeds security ownership in workload teams
- Centralize services for packages and dependencies
- Manual code reviews
- Automate testing throughout the development and release lifecycle
- Train for application security
- Regularly assess security properties of the pipelines
- Deploy software programmatically
- Perform regular penetration testing
-
-
Reliability
-
- How do you ensure sufficient gap between quotas and maximum usage to accommodate failover?
- How do you automate quota management?
- How do you monitor and manage service quotas?
- How do you accommodate fixed service quotas and constraints through architecture?
- How do you manage service quotas and constraints across accounts and Regions?
- How do you manage service quotas and constraints?
- How do you build a program that embeds reliability into workload teams?
-
- How do you enforce non-overlapping private IP address ranges in all private address spaces?
- How do you prefer hub-and-spoke topologies over many-to-many mesh?
- How do you ensure IP subnet allocation accounts for expansion and availability?
- How do you provision redundant connectivity between private networks in the cloud and on-premises environments?
- How do you use highly available network connectivity for workload public endpoints?
-
- Monitor end-to-end tracing of requests through your system
- Conduct reviews regularly
- Analytics
- Automate responses (Real-time processing and alarming)
- Send notifications (Real-time processing and alarming)
- Define and calculate metrics (Aggregation)
- Monitor End-to-End Tracing of Requests Through Your System
-
- Monitor all components of the workload to detect failures
- Fail over to healthy resources
- Automate healing on all layers
- Rely on the data plane and not the control plane during recovery
- Use static stability to prevent bimodal behavior
- Send notifications when events impact availability
- Architect your product to meet availability targets and uptime service level agreements (SLAs)
-
-
Cost Optimization
-
- Establish ownership of cost optimization
- Establish a partnership between finance and technology
- Establish cloud budgets and forecasts
- Implement cost awareness in your organizational processes
- Monitor cost proactively
- Keep up-to-date with new service releases
- Quantify business value from cost optimization
- Report and notify on cost optimization
- Create a cost-aware culture
-
- Perform cost analysis for different usage over time
- Analyze all components of this workload
- Perform a thorough analysis of each component
- Select components of this workload to optimize cost in line with organization priorities
- Perform cost analysis for different usage over time
- Select software with cost effective licensing
-
-
Performance
-
- Learn about and understand available cloud services and features
- Evaluate how trade-offs impact customers and architecture efficiency
- Use guidance from your cloud provider or an appropriate partner to learn about architecture patterns and best practices
- Factor cost into architectural decisions
- Use policies and reference architectures
- Use benchmarking to drive architectural decisions
- Use a data-driven approach for architectural choices
-
- Use purpose-built data store that best support your data access and storage requirements
- Collect and record data store performance metrics
- Evaluate available configuration options for data store
- Implement Strategies to Improve Query Performance in Data Store
- Implement data access patterns that utilize caching
-
- Understand how networking impacts performance
- Evaluate available networking features
- Choose appropriate dedicated connectivity or VPN for your workload
- Use load balancing to distribute traffic across multiple resources
- Choose network protocols to improve performance
- Choose your workload's location based on network requirements
- Optimize network configuration based on metrics
-
- Establish key performance indicators (KPIs) to measure workload health and performance
- Use monitoring solutions to understand the areas where performance is most critical
- Define a process to improve workload performance
- Review metrics at regular intervals
- Load test your workload
- Use automation to proactively remediate performance-related issues
- Keep your workload and services up-to-date
-
-
Sustainability
-
- Optimize geographic placement of workloads based on their networking requirements
- Align SLAs with sustainability goals
- Optimize geographic placement of workloads based on their networking requirements
- Stop the creation and maintenance of unused assets
- Optimize team member resources for activities performed
- Implement buffering or throttling to flatten the demand curve
-
- Optimize software and architecture for asynchronous and scheduled jobs
- Remove or refactor workload components with low or no use
- Optimize areas of code that consume the most time or resources
- Optimize impact on devices and equipment
- Use software patterns and architectures that best support data access and storage patterns
- Remove unneeded or redundant data
- Use technologies that support data access and storage patterns
- Use policies to manage the lifecycle of your datasets
- Use shared file systems or storage to access common data
- Back up data only when difficult to recreate
- Use elasticity and automation to expand block storage or file system
- Minimize data movement across networks
-
- Articles coming soon
< All Topics
Print
Implement distributed tracing
PostedNovember 6, 2024
UpdatedNovember 8, 2024
ByKevin McCaffrey
Implementing Distributed Tracing for Observability
Distributed tracing is a crucial tool for monitoring and visualizing requests as they move through different components of a distributed system. By capturing trace data from multiple sources and analyzing it in a unified view, teams gain deeper insights into request flows, identify bottlenecks, and pinpoint areas for optimization.
Capture End-to-End Trace Data
Instrument your application to capture trace data across all components involved in handling a request. This includes services, databases, external APIs, and any other components of your system. Capturing trace data from end to end provides a holistic view of how requests flow, helping teams understand the entire journey and detect where delays or failures may occur.
Use Unified View for Analysis
Consolidate trace data into a unified view to make it easier to analyze and understand how requests move through the system. A unified view enables teams to visualize the complete path of each request, identify which components are contributing to latency, and determine where improvements can be made to streamline request handling.
Identify Bottlenecks and Latency Issues
Use distributed tracing to identify bottlenecks, latency issues, and other performance problems within your system. By analyzing trace data, teams can see how long each component takes to process a request and determine which services are underperforming. Identifying bottlenecks helps prioritize optimization efforts where they are most needed.
Detect and Resolve Failures
Distributed tracing also helps in detecting failures within your system by pinpointing where a request fails or experiences an error. Trace data can show which component is responsible for a failure, allowing teams to quickly determine the root cause and resolve the issue before it impacts users. This capability is especially useful in complex, multi-service environments.
Optimize Inter-Service Communication
Use trace data to analyze how services interact with each other and optimize inter-service communication. Tracing can reveal inefficient communication patterns, such as redundant requests, unnecessary dependencies, or suboptimal routing. Optimizing these interactions helps improve the overall performance and reliability of the system.
Supporting Questions
- How is trace data captured across different components of the system?
- How is distributed tracing used to identify bottlenecks and latency issues?
- How does tracing help in detecting and resolving failures?
Roles and Responsibilities
Tracing Engineer
Responsibilities:
- Implement distributed tracing in the application to capture trace data across all system components.
- Ensure trace data is captured consistently and accurately to provide meaningful insights.
Performance Analyst
Responsibilities:
- Analyze trace data to identify bottlenecks and latency issues across the distributed system.
- Recommend optimizations based on trace data to improve request handling and inter-service communication.
Incident Responder
Responsibilities:
- Use trace data to detect and troubleshoot failures within the system.
- Resolve incidents quickly by determining the root cause using distributed tracing insights.
Artifacts
- Tracing Implementation Guide: A document outlining how distributed tracing is implemented across the system, including components being traced and data collection methods.
- Trace Analysis Dashboard: A visual representation of trace data, showing the request flow through different components, response times, and bottlenecks.
- Incident Resolution Log: A log capturing incidents detected through tracing, including actions taken and the outcome of those actions.
Relevant AWS Tools
Tracing and Monitoring Tools
- AWS X-Ray: Provides distributed tracing capabilities to capture and visualize the flow of requests through your application, helping to identify bottlenecks and performance issues.
- Amazon CloudWatch: Integrates with AWS X-Ray to provide metrics and alerts based on trace data, helping monitor system health and performance.
Logging and Visualization Tools
- Amazon CloudWatch Logs: Stores logs that complement trace data, providing additional context for understanding system behavior and troubleshooting issues.
- Amazon Managed Grafana: Visualizes trace data from AWS X-Ray, offering dashboards that help teams understand request flows and identify bottlenecks in real time.
Alerting Tools
- AWS SNS (Simple Notification Service): Sends notifications based on insights gathered from tracing, allowing teams to respond quickly to potential issues.
- AWS Lambda: Can be used to automate responses to specific tracing events, such as creating alerts or initiating failover processes when a bottleneck is detected.
Table of Contents