01 The challenge
Growth had been bought with bigger instances and more of them. Costs were rising faster than revenue, latency was creeping up, and the peak-season plan was simply to scale everything further — a plan that risked both the budget and the platform.
02 What we did
- Cost and performance audit across compute, storage, data transfer and managed services
- Right-sizing and autoscaling based on actual load patterns, with reserved and spot capacity where appropriate
- Application-level fixes: query optimisation, caching layers and asset delivery through a CDN
- Load testing at Black Friday multiples before the season
- Cost governance — budgets, tagging and alerts so spend stays visible
03 The results
- Monthly cloud spend cut by more than half
- Page load times improved substantially
- Black Friday traffic — many times normal — handled without emergency scaling
- Revenue grew roughly fivefold on a smaller infrastructure bill