Many inexperienced developers don't understand the difference in latency between a local memory call and a remote HTTP API call, so treat the two as equal in terms of architectural concerns.