Prometheus is free, Grafana is free, Loki is free—yet Datadog posted $2.3B in revenue and Shopify runs a 15-person team just to manage their observability stack. We decode which open source tools (Prometheus, Loki, Tempo, VictoriaMetrics) actually deliver on their promises, which hide massive operational complexity, and when the "free" option costs more than paying a vendor. Learn the decision framework that matches observability architecture to your team's operational maturity. In this episode:- Single-cluster Prometheus costs ~5 hrs/month ($750-1500 equivalent), but multi-cluster federation jumps to 40-80 hrs/month ($6K-12K)—know your tier before committing- Loki delivers 5-10x cheaper storage than OpenSearch but 3-5x slower queries for complex searches—works brilliantly for structured logs with good labels, struggles with full-text search- VictoriaMetrics reports 40-60% storage reduction vs Prometheus with better high-cardinality handling—consider it before jumping to commercial platforms Perfect for senior platform engineers, sres, devops engineers with 5+ years experience making build vs buy decisions for observability infrastructure looking to level up their platform engineering skills. Episode URL: https://platformengineeringplaybook.com/podcasts/00027-observability-tools-showdown