> This page is for version v0.7.0.
> For other versions, use one of these documentation indexes:
> - Latest (v1.5.1) (default): https://docs.nvidia.com/dynamo/latest/llms.txt
> - dev: https://docs.nvidia.com/dynamo/dev/llms.txt
> - v1.5.1: https://docs.nvidia.com/dynamo/v1.5.1/llms.txt
> - v1.5.0: https://docs.nvidia.com/dynamo/v1.5.0/llms.txt
> - v1.4.2: https://docs.nvidia.com/dynamo/v1.4.2/llms.txt
> - v1.4.1: https://docs.nvidia.com/dynamo/v1.4.1/llms.txt
> - v1.4.0: https://docs.nvidia.com/dynamo/v1.4.0/llms.txt
> - v1.3.0: https://docs.nvidia.com/dynamo/v1.3.0/llms.txt
> - v1.2.1: https://docs.nvidia.com/dynamo/v1.2.1/llms.txt
> - v1.2.0: https://docs.nvidia.com/dynamo/v1.2.0/llms.txt
> - v1.1.1: https://docs.nvidia.com/dynamo/v1.1.1/llms.txt
> - v1.1.0: https://docs.nvidia.com/dynamo/v1.1.0/llms.txt
> - v1.0.2: https://docs.nvidia.com/dynamo/v1.0.2/llms.txt
> - v1.0.1: https://docs.nvidia.com/dynamo/v1.0.1/llms.txt
> - v1.0.0: https://docs.nvidia.com/dynamo/v1.0.0/llms.txt
> - v0.9.1: https://docs.nvidia.com/dynamo/v-0-9-1/llms.txt
> - v0.9.0: https://docs.nvidia.com/dynamo/v-0-9-0/llms.txt
> - v0.8.1: https://docs.nvidia.com/dynamo/v-0-8-1/llms.txt
> - v0.8.0: https://docs.nvidia.com/dynamo/v-0-8-0/llms.txt
> - v0.7.1: https://docs.nvidia.com/dynamo/v-0-7-1/llms.txt
> - v0.7.0: https://docs.nvidia.com/dynamo/v-0-7-0/llms.txt

> For clean Markdown content of this page, append .md to this URL. For the complete documentation index, see https://docs.nvidia.com/dynamo/llms.txt. For full content including API reference and SDK examples, see https://docs.nvidia.com/dynamo/llms-full.txt.

# KV Cache Transfer in Disaggregated Serving

In disaggregated serving architectures, KV cache must be transferred between prefill and decode workers. TensorRT-LLM supports two methods for this transfer:

## Default Method: UCX
By default, TensorRT-LLM uses UCX (Unified Communication X) for KV cache transfer between prefill and decode workers. UCX provides high-performance communication optimized for GPU-to-GPU transfers.

## Beta Method: NIXL
TensorRT-LLM also supports using **NIXL** (NVIDIA Inference Xfer Library) for KV cache transfer. [NIXL](https://github.com/ai-dynamo/nixl) is NVIDIA's high-performance communication library designed for efficient data transfer in distributed GPU environments.

**Note:** NIXL support in TensorRT-LLM is currently beta and may have some sharp edges.

## Using NIXL for KV Cache Transfer

**Note:** NIXL version shipped with current dynamo is not supported by tensorrt-llm<=1.2.0rc2. In order to use NIXL backend for KV cache transfer, users are required to build container image with tensorrt-llm>=1.2.0rc3.

To enable NIXL for KV cache transfer in disaggregated serving:

1. **Build the container with NIXL support(tensorrt-llm==1.2.0rc3):**
   ```bash
   ./container/build.sh --framework trtllm \
     --tensorrtllm-pip-wheel tensorrt-llm==1.2.0rc3
   ```

2. **Run the containerized environment:**
   See [run container](/dynamo/v-0-7-0/components/backends/tensor-rt-llm#run-container) section to learn how to start the container image built in previous step.

   Within container, unset `TRTLLM_USE_UCX_KVCACHE` variable so NIXL can be used instead of UCX.

   ```bash
    unset TRTLLM_USE_UCX_KVCACHE
    ```

3. **Start the disaggregated service:**
   See [disaggregated serving](/dynamo/v-0-7-0/components/backends/tensor-rt-llm#disaggregated-serving) to see how to start the deployment.

4. **Send the request:**
   See [client](/dynamo/v-0-7-0/components/backends/tensor-rt-llm#client) section to learn how to send the request to deployment.

**Important:** Ensure that ETCD and NATS services are running before starting the service.