🌐 CDN Edge Caching Bandwidth Offload
90% cache hit ratio means 10% of traffic still hits origin. At 10M requests/month with 50 KB average response, that is 50 GB of origin egress. At $0.09/GB, that is $4.50. Not scary. At 1B requests with 500 KB average, it is $45,000/month. Model it before you budget.
📦 Cache Configuration
Requests/month × average response size × cache-hit split. Total traffic, CDN-offloaded GB, residual origin traffic, and the dollar cost of what the cache does not absorb. Four numbers that anchor your CDN procurement conversation.
CDN Cache Hit Ratio: 90% Caching = 10× Origin Cost Reduction
Content Delivery Networks (CDNs) operate on a simple but powerful premise: serve cached content from edge locations close to users, and only forward cache-miss requests to the origin server. Each percentage point of cache hit ratio directly reduces origin egress traffic, server load, and infrastructure cost. At a 90% cache hit ratio, only 10% of total traffic touches the origin — a 10× reduction in origin bandwidth demand and associated egress charges.
How This Calculator Models Cache Offload
The calculation is straightforward yet often overlooked during architecture planning:
- Total Traffic (GB) = Total Requests × Average Response Size (KB) ÷ 1,000,000
- CDN Offloaded (GB) = Total Traffic × Cache Hit Ratio
- Origin Traffic (GB) = Total Traffic − CDN Offloaded
- Origin Egress Cost ($) = Origin Traffic × Per-GB Egress Rate
Note: this calculator uses a flat per-GB rate for origin egress. For AWS tiered pricing (which applies volume discounts at scale), use the AWS Egress Cost Estimator for a more precise multi-band calculation.
Cache Hit Ratio Benchmarks by Content Type
| Content Type | Typical Cache Hit Ratio | Key Factors |
|---|---|---|
| Static Assets (CSS, JS, images, fonts) | 95–99% | Versioned URLs (fingerprinted filenames) enable indefinite caching. Immutable assets hit cache on every request after the first. |
| API Responses (semi-static data) | 80–95% | Product catalogs, configuration, and reference data change infrequently. Set short TTLs (60–300s) with stale-while-revalidate to balance freshness and cache efficiency. |
| User-Generated Content (profiles, posts) | 50–80% | Popular content is read frequently; tail content is rarely accessed. Cache warming and selective TTLs help optimize the Pareto distribution of requests. |
| Personalized / Authenticated Responses | 0–20% | Per-user data is difficult to cache at the edge. Edge-side includes (ESI) or client-side personalization can separate cached templates from dynamic data. |
CloudFront Free Tier: The Hidden Cost Advantage
AWS CloudFront includes a generous free tier: 1 TB of data transfer out per month, 10 million HTTP/HTTPS requests per month, and 2 million CloudFront Function invocations — all free, every month, indefinitely. For workloads under these limits, the CDN layer is effectively free, making the origin egress savings pure profit. Even beyond the free tier, CloudFront egress ($0.085/GB in the US) is often cheaper than direct S3/EC2 egress ($0.09/GB) for the first 10 TB — and the cache hit ratio multiplies those savings by reducing origin traffic.
Edge Caching Strategies for Maximum Offload
Maximizing cache hit ratio requires deliberate engineering. Use content-addressable URLs (hash in filename) for immutable assets to enable long-term caching. Apply Cache-Control: public, max-age=31536000, immutable for fingerprinted assets. For dynamic content, implement stale-while-revalidate (RFC 5861) to serve stale cached content while asynchronously refreshing — users never wait for origin latency. Configure origin shield (regional cache layers in CloudFront) to collapse multiple concurrent cache-miss requests into a single origin fetch, reducing origin load by an additional 60-80% for popular cacheable content.