Components

Router

View as Markdown

The Dynamo KV Router intelligently routes requests by evaluating their computational costs across different workers. It considers both decoding costs (from active blocks) and prefill costs (from newly computed blocks), using KV cache overlap to minimize redundant computation. Optimizing the KV Router is critical for achieving maximum throughput and minimum latency in distributed inference setups.

Quick Start

To launch the Dynamo frontend with the KV Router:

$python -m dynamo.frontend --router-mode kv --http-port 8000

For Kubernetes, set DYN_ROUTER_MODE=kv on the Frontend service. Workers automatically report KV cache events — no worker-side configuration changes needed.

ArgumentDefaultDescription
--router-mode kvround_robinEnable KV cache-aware routing
--router-kv-overlap-score-weight1.0Balance prefill vs decode optimization (higher = better TTFT)
--no-router-kv-eventsenabledFall back to approximate routing (no event consumption from workers)
--router-queue-threshold4.0Backpressure queue threshold; enables priority scheduling via nvext.agent_hints.priority
--router-queue-policyfcfsQueue scheduling policy: fcfs (tail TTFT), wspt (avg TTFT), or lcfs (comparison-only reverse ordering)
--no-router-track-prefill-tokensdisabledIgnore prompt-side prefill tokens in router load accounting; useful for decode-only routing paths

Standalone Router

You can also run the KV router as a standalone service (without the Dynamo frontend). See the Standalone Router component for more details.

For deployment modes and quick start steps, see the Router Guide. For CLI arguments and tuning guidelines, see Configuration and Tuning. For A/B benchmarking, see the KV Router A/B Benchmarking Guide.

Prerequisites and Limitations

Requirements:

  • Dynamic endpoints only: KV router requires register_model() with model_input=ModelInput.Tokens. Your backend handler receives pre-tokenized requests with token_ids instead of raw text.
  • Backend workers must call register_model() with model_input=ModelInput.Tokens (see Backend Guide)
  • You cannot use --static-endpoint mode with KV routing (use dynamic discovery instead)

Multimodal Support:

  • TRT-LLM and vLLM: Multimodal routing supported for images via multimodal hashes
  • SGLang: Image routing not yet supported
  • Other modalities (audio, video, etc.): Not yet supported

Limitations:

  • Static endpoints not supported—KV router requires dynamic model discovery via etcd to track worker instances and their KV cache states

For basic model registration without KV routing, use --router-mode round-robin, --router-mode random, --router-mode least-loaded, or --router-mode device-aware-weighted with both static and dynamic endpoints.

Next Steps