200+ subdomains from Akamai → AWS in 1.5 months — zero downtime
Context
- High‑traffic B2C brand (millions of daily requests) across India & SEA
- 200+ subdomains with varied cache keys, signed URLs, device/CDN behaviors
- Heavy Akamai feature usage; rising egress & CDN costs
Problem
- Performance variability in key regions; limited IaC; vendor lock‑in risk
- Rules drift and inconsistent cache policies across subdomains
Solution
- Parity map Akamai behaviors → CloudFront (origins, cache keys, headers)
- Edge function rewrite: EdgeWorkers → Lambda@Edge
- WAF: COUNT → tune → BLOCK; bot mgmt mapping; rate limits
- Cache warming, RUM baseline; blue/green DNS & real‑time dashboards
Outcome
- Cutover with 0 downtime (windows < 30min)
- 35% egress/CDN cost reduction
- Cache hit ratio +12% (to 98–99%)
- Delivery time p95 improved by 14% in India
- 1.Assess
Inventory, configs, traffic profile, risks
- 2.Plan
Akamai→AWS parity map, rollback paths, KPIs
- 3.Simulate
Lab + shadow traffic; WAF COUNT mode
- 4.Dual‑Run
Mirrored paths, compare logs/latency, tune rules
- 5.Cutover
Cohorts; time‑boxed windows; live dashboards
- 6.Optimize
Cache keys, TLS, Brotli/GZIP, cost tuning
We’ll map parity, simulate with shadow traffic, then dual‑run before cutover.
Talk to an expert