C4: Container, Code, Cloud & Context

Observability in serverless is challenging. Lambda functions execute in ephemeral environments, making traditional APM tools ineffective. AWS Lambda Powertools for .NET provides structured logging, distributed tracing, and custom metrics out of the box—implementing AWS Well-Architected best practices without boilerplate. This guide covers installation, configuration, and production patterns for each Powertools utility. The Observability Challenge in […]

GitHub Copilot is now Generally Available. For enterprises, adoption requires addressing IP concerns, security, and developer training. Key Settings Suggestions Matching Public Code: Disable to avoid copyright concerns. Allow Suggestions from Public Repos: Enable for better accuracy (toggleable). Productivity Impact In our pilot, developers reported 30-40% reduction in boilerplate time. However, code review time increased […]

At scale, telemetry volume overwhelms budgets. Sampling reduces data while preserving insights. Adaptive Sampling (Default) Automatically adjusts rate to stay within 5 items/second per server. Good for most cases. Fixed-Rate Sampling Tip: Exclude failures and slow requests from sampling to ensure you never miss issues.

CloudWatch Logs Insights uses a SQL-like query language. Here are my most-used patterns. Find Errors Latency Percentiles

Cosmos DB now supports multi-level partition keys. This solves the “hot partition” problem for multi-tenant applications. Data is now partitioned by Tenant → Year → Order. This distributes write load across many physical partitions while allowing efficient queries within a tenant.

Single Table Design (STD) is the most controversial and misunderstood DynamoDB pattern. The premise: store all entity types in one table, use generic partition and sort keys, and leverage Composite Primary Keys and GSIs for access patterns. When done correctly, STD eliminates N+1 queries, reduces costs, and simplifies IAM. When done incorrectly, it creates debugging […]