Cost optimization
summary
Tesco finds £1.8M of waste inside a £3M cloud bill and actually cuts about £600k by consolidating idle databases, right‑sizing AKS (clusters running at ~10% utilization) and switching off test instances on nights and weekends. Observability uses Prometheus/VictoriaMetrics, Grafana, Splunk and New Relic with distributed tracing, and an open source deployer called Aqueduct to push images efficiently across tills. Hybrid decisions balance public scale for peak season against private cloud costs of roughly one‑fifth to one‑eighth. shared services hybrid cloud switch off test systems
AI-written summary · about
chapters (24)
- 0:00Cost optimization intro
- 3:34Why cost matters in SDLC
- 6:06Underutilized resources waste
- 8:38Access patterns and test systems
- 10:06Shared services benefits and challenges
- 13:16Public vs private cloud trade-offs
- 15:47Workload placement and traffic costs
- 16:51Hidden costs in development
- 19:16Tesco edge architecture
- 21:04Resource constraints and IOPS
- 22:51Shared components governance
- 24:18Observability in edge environments
- 26:26Code pollution from logging
- 27:27Distributed deployment and flags
- 29:33ML for pricing and cost control
- 32:22No silver bullet for costs
- 33:48Measuring hidden costs
- 35:36Hard-to-replicate cloud services
- 37:44Fixing code pollution in metrics
- 39:09Application performance monitoring stack
- 40:35600k savings and low-hanging fruits
- 43:06Disk storage optimization
- 44:51Removing unnecessary metrics
- 45:11Public to private cloud savings
related talks
41mFinOps and Cloud Cost Optimization
23mFinOps and Cloud Cost Optimization
34mOptimizing Cloud Efficiency with Prompt Engineering
46mWhen the CFO Comes Knocking: Our Journey to a Cost Efficient Data Platform
31mThe irony of cloud cost-cutting: when saving money leads to losing money
49mBuilding a zero-waste compute infrastructure on
36mFinOps & Observability: technical implementation
05mCutting 2/3rds of your cloud costs with common sense