How do you plan for IPL, Diwali, and Ganesh Chaturthi peak load?

Mumbai retail has four predictable peaks: IPL Apr, May (entertainment, food delivery, alcohol-adjacent D2C, Wankhede match-day surges), Ganesh Chaturthi Aug, Sep (Maharashtra-specific retail, idol + decoration commerce), Diwali Oct, Nov (the big one, 4-8× baseline traffic + GMV), and wedding Nov, Feb (Worli-Lower-Parel luxury, jewellery, sarees, decor). We treat each as a separate release-train. Plan template:

• T-60 days: load-test at 5× baseline on staging, fix N+1 queries, warm Varnish, profile slow EAV joins.
• T-30 days: feature freeze. Read-replica MySQL added, OpenSearch reindexed, Redis cluster scaled, image CDN warmed.
• T-7 days: deploy freeze, war-room calendar locked, on-call rota set across India + Ahmedabad team.
• Peak day: 24/7 IST war-room, Cloudflare + Varnish full-page cache, EC2 auto-scale on, observability via NewRelic / Datadog, real-time order-error Slack alerts.
• T+7 days: post-mortem with metrics, code freeze lifted, lessons captured for next year’s playbook.

Standard tier includes one peak war-room; Enterprise tier includes annual coverage across all four Mumbai peaks.