CDN Request Cost Calculator

What to split into separate scenarios (request classes)

• Static vs dynamic: some CDNs price them differently.
• Cache hit vs miss: misses also create origin load and origin egress.
• Regions: if pricing varies by geography, model your top regions separately and add them up.

Use the Pricing unit selector in the calculator to enter $/10k or $/1M directly (it keeps the same effective rate).

When request fees spike

• Bot traffic: more requests with little bandwidth change.
• Chatty clients: many small API calls per page view.
• Cache busting: query-string churn and low TTLs can increase request volume.

Cache keys and TTL levers

• Query string and header variations can multiply unique cache keys.
• Short TTLs increase origin fetches and raise request volume.
• Normalize URLs to reduce cache fragmentation when possible.

Worked request estimate from pages and assets

• Page views/month x assets per page = asset requests.
• Add API or dynamic calls as a separate request class.
• Apply a bot or prefetch multiplier if relevant.

CDN request anatomy

• Image variants (size, format) create new cache keys.
• HEAD requests and health checks still count as requests.
• Range requests for media can multiply request counts.

CDN request cost is mostly a request-shape and cache-key problem

This page should not be treated like a simple rate converter. CDN request fees usually become interesting when request classes fragment, cache keys multiply, or bots and client behavior add requests without adding much bandwidth.

• Edge analytics: the real source for billable request counts, not origin-side counts.
• Request classes: static, dynamic, hit, miss, or other classes may need separate assumptions.
• Cache-key behavior: headers, query strings, and URL variations can raise request volume faster than traffic growth alone.

Where CDN request estimates usually drift

• Page views look stable, but assets per page, API fan-out, and image variants increase total edge requests.
• Bot traffic, health checks, and cache-busting behavior raise requests without moving delivered GB very much.
• Teams use one request class even though dynamic or miss traffic is behaving differently from the main static path.
• Request growth is blamed on users when the real issue is cache-key fragmentation or low TTL behavior.

What to review before trusting the request baseline

• Use edge request counts and split high-noise traffic classes if pricing or behavior differs.
• Check assets per page, image variant generation, and API fan-out before assuming user growth caused the change.
• Review query-string churn, header-based cache keys, and short TTLs because they often create the real multiplier.
• Treat bot and incident traffic as a separate scenario instead of hiding it in one average month.

Baseline vs bot-heavy request scenarios

How to review the first real CDN request month

• Compare billed request units to edge analytics and identify whether the miss came from class mix, bot spikes, or cache-key sprawl.
• Watch request volume and cache hit rate together so traffic-quality changes do not get mistaken for normal growth.

Next steps