In many IO-heavy enrichments we use cache (LRU mostly):
Weather enrichment
Currency conversion
IP Lookups
API Request
SQL Query
~Referer parser enrichment~ (loads everything into memory straight away, as well as UA parser)
I'm wondering how many cache misses do we usually have, how much the caches used in general. Knowing that we can decide if we should execute those enrichments on separate thread pool (if still IO-heavy - separate blocking thread pool, if cache used up to 99% - main fixed thread pool).
In many IO-heavy enrichments we use cache (LRU mostly):
I'm wondering how many cache misses do we usually have, how much the caches used in general. Knowing that we can decide if we should execute those enrichments on separate thread pool (if still IO-heavy - separate blocking thread pool, if cache used up to 99% - main fixed thread pool).