GraphCDN's automatic mutation invalidation, cache purging, and fine-grained configuration ensure our customers get the highest cache hit rate possible. We usually see cache hit rates ranging from 50% to 95% depending on the use case.
There are some reasons beyond the configuration and purging that affect the cache hit rate:
- Authenticated users: Query results that differ between authenticated user should be cached separately with scopes.
- Multiple locations: Every one of GraphCDN's 60 worldwide locations has its own local cache.
- Query size and other limits: If you respond with large amounts of data (133mb+), we might not be able to cache it.
- LRU caching: Finally, we use an LRU (least-recently-used) cache, so uncommonly requested data might be evicted before the maxAge or swr time expires.
If necessary, contact us at firstname.lastname@example.org to talk through specific strategies to optimize your cache hit rate.