Tag: #라우팅
Kubernetes, DevOps and MLOps, written from things we ran and measured ourselves · 1 posts
Hands-On with NVIDIA Dynamo: KV-Aware Routing Got Almost Everything Right, but Disaggregated Serving Was 5x Slower on a Single GPU and Over 20x Slower Across Two Servers
We installed NVIDIA Dynamo on a single-GPU environment without Kubernetes and measured frontend overhead, KV-aware routing hit rates, silent failures when routers fail to receive events, the overhead of prefill/decode di
2026-10-07 · 23 min read #Dynamo#LLM서빙#KV캐시#라우팅#prefill-decode분리