Delivery
How caching works
The proxy caches eligible responses in fast memory storage and mirrors each record in the database, so cache state is quick to read at request time and still visible from the dashboard. This topic explains what is eligible, how long entries live, and where the resulting hit ratio shows up on the Cache screen.
How it works
- Only public, cacheable GET/HEAD responses under 1 MB are eligible
- TTL comes from the origin's max-age header, clamped to platform limits rather than trusted outright
- Cache hit ratio is one of the headline metrics on the dashboard overview
Good to know
- A database record mirrors the cached payload for reporting and purge lookups, unique per tenant and cache key
- Responses carrying Set-Cookie or private cache directives are never stored
- Non-GET/HEAD methods always bypass the cache
- Changing a cache rule requires a deploy before it affects live traffic