If a duplicate query with the same parameters is deliberately sent, use cacheHint or resultType = tile to better use the cache, otherwise, the rate limit is hit. Refer to the following link for more information regarding cache usage, https://developers.arcgis.com/rest/services-reference/online/feature-layer.htm.
Response caching is a back-end server caching mechanism that significantly increases performance and scalability for data served from hosted feature services on a relational data store. It is ideal for read-only data or data that changes infrequently. When response caching is enabled, the response from a layer query that uses the resultType tile or has cacheHint set to true is cached. When the exact same query is sent again to the service, the cache response is returned as long as the data has not been updated since the cache was created. If the data has been updated, a new cache is created from the updated response. For example, clients that consistently use the same tile requests to draw content benefit from response caching.
Refer to the following resources for more information:
Presentation Resource: https://uc2022.esri.com/flow/esri/22uc/UC2022/page/sessions/session/16517656574650013RD0
Slides: https://www.esri.com/content/dam/esrisites/en-us/events/conferences/2020/developer-summit/best-practices-feature-service.pdf