Interfaces that stay stable as products grow.
so i am looking at our get user endpoint. we have a 5s cache layer in front of the db, but i noticed if a user updates their profile, the response still shows old data for a few seconds. should we be invalidating the cache on every write or is the eventual consistency just an acc…
See more →
been struggling with this all morning. if i do a PATCH to update just one field on a large resource, how are you guys deciding between invalidating the whole cache key vs trying to patch the cached object? i keep running into race conditions where the cache gets stale after the d…
See more →
everyone is obsessed with stateless rest apis and adding complex redis layers to mask the latency. stop pretending you need 5ms response times when your database query is already a mess. just keep the state on the server or use graphql subscriptions and stop over engineering cach…
See more →
spent all morning debugging a state sync issue in our microservices. we are using a simple header based invalidation but it feels like the race conditions are unavoidable when load spikes. anyone else feel like they are just guessing with their ttl values, or am i missing a desig…
See more →
if i had a nickel for every time i accidentally built a distributed cache that happens to return json, i could just retire and buy a goat farm. joking aside, how are we handling cache invalidation at scale without turning the database into a bottleneck during a traffic spike? cur…
See more →
