Practical guide
Use cache ratio to forecast Cloud CDN cost
Turn Cloud CDN delivery volume and cache-hit ratio into separate cache-fill and origin-cost assumptions.
Updated · Sources checked
Hit ratio is an input, not a saving percentage
Cloud CDN charges delivery and lookup on cacheable hits; cacheable misses add cache fill and applicable origin charges. A hit ratio changes which bytes reach the origin but does not make delivery free.
For 20 TiB delivered, an 80% hit ratio produces a simple initial assumption of 16 TiB hits and 4 TiB fills. At 95%, it becomes 19 TiB and 1 TiB. Compare both with measured cache metrics, then price cache transfer, lookup requests, fill source/destination, and origin processing using current tables. Headers, bypass paths, and object changes can make the simple fill assumption imperfect.
Improve cacheability only where content correctness permits: stable static assets, appropriate cache headers, and versioned URLs are common candidates. Avoid forcing private or rapidly changing responses into a cache policy solely for cost. Use the Cloud CDN Cost Calculator for scenarios.
FAQ
Does cache fill have one global rate?
No. Google documents fill pricing by source and destination relationship.
Should I forecast request count?
Yes. Lookup requests are a separate metered quantity.