Give me 2 minutes, and I'll teach you how Disney+ Hotstar scaled to 25 million concurrent users (not joking):
• They ran tests to find bottlenecks
• They used a multi-level cache to scale reads
• They rate limited the traffic to prevent server overload
• They used baked container images for faster provisioning
• They used different instance types to get extra servers from AWS
• They installed panic mode on the server to avoid server overload
• They focused only on the high availability of important components
• They added jitter and caching to prevent the thundering herd problem
• They kept an open communication channel to align on important metrics
• They pre-warmed the infrastructure ahead of major events by estimating peak traffic