Benchmark & decision framework for KV cache transfer compression in disaggregated LLM serving — 14+ compressors, 6 models, GPU-calibrated timing, decision flowchart

A fault-tolerant LLM routing system that decouples inference from AWS Bedrock by routing prefill and decode tasks through SQS and ensuring zero-downtime scaling with a graceful drain sidecar.