Building Event-Driven Services - past events relevance

This question is a follow up after watching the Building Event-Driven Services webinar.
After understanding the general concept of event driven microservices with usage of Kafka, I am wondering how considered best to handle a service that takes under account past event information. For example a payment processing service that is driven by “payment happened” events but needs to check for some reason if that user made a related payment in the past few hours.
Ideally a local caching of events info is possible, however in the case where only limited caching is possible and event traffic is very high, would it be standard to consider caching of relevant data in some sort of quick access DB such as redis?

Thank you.