Caching Metrics

At Stellate, we offer a few metrics related to caching. You can use these to understand and optimize your cache performance. Insights from these metrics allow you to improve your cache hits, as a result of that, reduce the traffic to your backend.

Cache Hit Rate (CHR)

The CHR metric indicates a ratio of hits to cacheable traffic. We calculate CHR using the following formula:

Cache Hit Rate calculation formula, which is equal to the number of cache hits divided by the sum of cache hits, cache misses, and cacheable passes.

Max CHR

The Max CHR is the potential maximum Cache Hit Rate achievable by your service. We calculate this by looking at your traffic, and then computing the different combinations of GraphQL variables, responses, and a few other fields that affect our caching, and then computing how many different caching buckets you would need, and with this we can get that maximum percentage.

Although achieving this maximum CHR might be challenging or, in some cases, practically impossible, it provides a valuable aspirational metric. It can help you understand how much your CHR can be improved.

Max Cache Hit Rate calculation formula, which is equal to the number of cache hits divided by the sum of cache hits, cache misses, and cacheable passes.

Uncached bandwidth

This metric tells you how much of the traffic that goes to Stellate ends up going to your backend, the more traffic you cache the smaller this number will get. It can help estimate bandwidth cost savings, as cached traffic doesn't impact your backend.

CHR Impact

This metric will tell you what is the percentage gain from optimizing a specific operation CHR. This is like saying “increasing the getUsers query will yield at most 11% in the overall service cache hit rate”.

You can see this metric inside of the Caching metrics page in your Stellate's dashboard, and you can use it to understand which operations are worth optimizing.