Amazon's finops: Cloud cost lessons from a global e-commerce giant (AMZ308)

Session Notes

Key Lessons from Amazon's Modernization Journey

Building on AWS Billing and Cost Management Services
Foundation for FinOps practices
Amazon operates on AWS, necessitating comprehensive FinOps stance
Started with custom financial reporting for monthly cloud costs
Transitioning to AWS data exports, cost and usage reports (CUR), and other AWS billing services
Moving from monthly/account grain to ARN and hourly grain visualization
Democratizing cost data across teams (builders, leaders, finance, FinOps)
Enabling Cost Explorer for self-service cost analysis and real-time decision making
Deploying organization-wide tagging strategies for better cost controls
Integrating with AWS features like Compute Optimizer and Cost Optimization Hub
Shifting from centralized reporting to a distributed model of cost intelligence
Leveraging AWS Organizations for consistent controls and aggregated tool set

Driving Efficiency Through Business-Aligned Mechanisms

Challenge: widespread adoption of cost management practices
Key insight: connect cloud costs to business outcomes at the team level
Combining granular costs from AWS Cost & Usage Report (CUR) with business metrics teams care about
Allows visibility into spending and value received per dollar spent

Scaling FinOps Practice Through Intelligent Automation

Focus on automating processes to scale FinOps efficiently
Details on specific automation strategies and tools used
Emphasizing the importance of automation in maintaining and scaling FinOps practices

Key to adoption: connecting costs to business outcomes for teams
Included accounts and tag-based cost allocation to business context and investment tracking
Automated return on investment analysis with AWS cost management services
Simplifying Cost Visibility
Used AWS Cloud Intelligence Dashboard (CID) for actionable insights and optimization opportunities
Nested AWS service-specific budget data against actual usage to show budgetary variances
Enabled finance and operational efficiency teams to address cost reduction initiatives and budget/forecast revisions
Integrated contextualized business data alongside AWS infrastructure usage
Created role-specific views for in-depth service analysis by teams
Real-World Example
Teams can identify cost spikes, origin accounts, services, and resources
Visualize business impact alongside metrics like revenue or budget
Tie cost spikes to individuals or initiatives for immediate action
Evolving Approach to Efficiency
Recognized efficiency is not one-size-fits-all
Measured basic resource utilization metrics (CPU, memory, network throughput)
Different workloads required different efficiency approaches
Built central efficiency mechanisms
Tracked business-specific efficiency
Monitored resource utilization against centrally agreed-upon ideals
Achieved significant gains in efficiency and cost reduction for applicable lines of business

Created a metric called the credit score to measure resource efficiency across various services
Iterative approach to hone central baseline
Evaluated alignment with central efficiency campaigns (e.g., capacity utilization, storage class optimization)
Correlated with business data (revenue, budget) to measure FinOps maturity across lines of business
Enabled teams to optimize and save significant amounts of money
Weekly Efficiency Score
Teams received a weekly efficiency score with cost recommendations
Recommendations could be grouped by technology category (storage, compute, generative AI, database, network)
Recommendations associated with accounts, teams, and owners
All stakeholders (finance teams, leadership, technology owners, operational efficiency teams) could view data through their respective lenses
Automation for Scaling FinOps Practices
True cloud financial management is a continuous intelligent cycle
Integrated AWS services with automated workflows to transform manual processes into self-improving systems
Journey towards automating FinOps started with providing better visibility into cloud costs
Continuous iteration to create systems with deeper insights into infrastructure spending patterns
Each improvement feeds back into the learning cycle, enhancing capabilities towards intelligent cloud financial management
Effective FinOps automation requires comprehensive usage of trends, budget variance, and capacity requirements

Teams receive notifications of optimization opportunities through preferred channels
For well-understood scenarios, teams can define policies and thresholds that trigger automated responses
Balance between automation and oversight is crucial for building trust and driving adoption
Building Trust Through Transparency
Scaling FinOps requires consistent transparency
Every action (automated or manual) must be logged and tracked in detail
Teams need to see exactly what's happening with their infrastructure costs and why
Transparency-first approach is crucial for adoption of automation and planning capabilities
Combining Human Insight with Automated Analysis
Particularly impactful in retail
Start FinOps automation journey with AWS services as the foundation
Focus first on gaining visibility into costs
Gradually automate well-understood processes (e.g., financial planning, OP1 cycle, planning, capacity management)
As trust is built through transparency and results, expand the scope and sophistication of automation
Goal is not to remove humans, but to enhance their capabilities with automation

Start with custom tools, mechanisms, and processes
AWS billing and cost management services provide better visibility than custom solutions
AWS Cost and Usage Report offers granular data for fine-grain cost control
Cost Explorer provides analysis capabilities directly to teams
AWS Organizations enables governance at massive scale
Connecting Business Outcomes to Cloud Costs
Real transformation occurs when business outcomes are connected to cloud costs
Understanding cost attribution to revenue is more actionable than just knowing spend
AWS Cost Intelligence Dashboards show team-level value and metrics aligned with goals
Automation for Scale and Efficiency
Move from monthly reviews to daily optimization actions
Intelligent systems detect anomalies, recommend optimization, and implement improvements automatically
Teams focus on strategic decisions while automation handles routine tasks
Operating at scale with automation