Performance & Speed Playbook

Goal

• Sub-second TTFB
• High cache-hit ratio
• Stable p95 latency under load

Dynamic webpages caching

• Cache dynamic HTML for key templates like Home, PDP, and PLP
• Use query-string-aware cache keys per URL and query
• Cache safely - avoid caching personalized or session-dependent responses

Edge delivery and optimization

• Edge cache static assets aggressively to reduce origin dependency and absorb traffic spikes
• Optimize images on the fly - resize and compress, and convert to modern formats when supported
• Use a global CDN footprint to reduce latency and improve resilience
• Use Anycast authoritative DNS for fast resolution and DDoS-resilient nameservers
• Keep origin hosting optimized for the web server and database, with CDN offload to minimize origin load

Database performance

Structure

• Choose clustered primary keys carefully
• Avoid wide clustered keys
• Keep a normalized core model
• Use pre-aggregated or materialized tables for heavy read patterns

Indexes

• Maintain optimized indexes
• Run periodic index maintenance
• Use computed and persisted columns where helpful
• Use full-text catalogs for search where relevant

Queries

• Use parameterized SQL to avoid ad-hoc plan bloat
• Batch writes
• Paginate reads
• Avoid RBAR (row-by-agonizing-row)

Block unwanted traffic

• Block traffic by country or region at the edge to reduce bot and scanner traffic before it reaches the site
• Block suspicious IP addresses and ranges at the web server
• Block known bad bots via User-Agent pattern rules at the web server level