Just heard about Prometheus with a web search. Wondering if there are use cases where this tech is used as the backend for a rate computing or rate limiting service for a very high throughput service. Let’s say I want to compute rate of a service that can go up to 250K Requests per second. The rate can increase from 20K RPS to 40 RPS in 2 minutes and similarly can drop also. I want to be able to react to it.
Thanks!
Do you mean auto-scaling? This is commonly done in Kubernetes using KEDA auto-scaling controller.
When people talk about rate limiting, we usually talk about denying requests over a specific rate. Things like “No more than 10 requests per second from each IP address”. This is not something Prometheus would do, as the cardinality tracking would be too high. You need specialized software for this. For example Fastly edge rate limits.