the real bottleneck isnt ur network bandwidth, its the browser trying to parse massive json payloads. once u hit hundreds of megabytes, the main thread just
dies while buffering bytes and managing object graphs. instead of cranking up timeouts, we started moving away from synchronous patterns entirely. we integrated
temporal and kafka to handle the heavy lifting in the background b4 it ever hits the client.
it turns out RAG is actually great for filtering this data down to something usable .
>the goal is to stop treating huge responses as single events. has anyone else found success moving toward an asynchronous architecture for large datasets?more here:
https://dzone.com/articles/large-api-responses-client