Included in the NGC Helm Repository is a chart designed to automate for push-button deployment to a Kubernetes cluster.
The Riva AI Services Helm Chart can be used to deploy ASR, NLP, and TTS services automatically. The Helm chart performs a number of functions:
Pulls Docker images from NGC for the Riva Speech Server, and utility containers for downloading and converting models.
Downloads the requested model artifacts from NGC as configured in the
Generates the Triton Inference Server model repository.
Starts Riva Speech as configured in a Kubernetes pod.
Exposes the Riva Speech Server as configured services.
Example pre-trained models are released with Riva for each of the services. The Helm chart comes pre-configured for downloading and deploying all of these models.
The Helm chart configuration can be modified for your use case by modifying
values.yaml file. In this file, you can change the settings related
to which models to deploy, where to store them, and how to expose the services.
Validate Kubernetes with NVIDIA GPU support.
Kubernetes with GPU is well supported by NVIDIA. The instructions at Install Kubernetes should be consulted to ensure your environment is properly setup.
If using an NVIDIA A100 GPU with Multi-Instance GPU (MIG) support, refer to MIG Support in Kubernetes.
Download and modify the Helm chart for your use. Fetch it from NGC:
export NGC_API_KEY=<your_api_key> helm fetch https://helm.ngc.nvidia.com/nvidia/riva/charts/riva-api-1.8.0-beta.tgz \ --username=\$oauthtoken --password=$NGC_API_KEY --untar
The above comment creates a new directory called
riva-apiin your current working directory. Within that directory is a
values.yamlfile which can be modified to suit your use case (see kubernetes_secrets and Riva Settings).
values.yamlfile has been updated to reflect the deployment requirements, Riva can be deployed to the Kubernetes cluster:
helm install riva-api riva-api
Alternatively, use the
--setoption to install without modifying the
values.yamlfile. Ensure you set the NGC API key, email, and
model_key_stringto the appropriate values. By default,
helm install riva-api \ --set ngcCredentials.password=`echo -n $NGC_API_KEY | base64 -w0` \ --set ngcCredentials.email=your_email@your_domain.com \ --set modelRepoGenerator.modelDeployKey=`echo -n model_key_string | base64 -w0`
Helm configuration. The following sections point out a few key areas of the
values.yamlfile and considerations for deployment. Consult the individual service documentation for more details as well as the Helm chart’s
values.yamlfile, which contains inline comments explaining the configuration options.
To deploy Riva, a functioning Kubernetes environment with a GPU (NVIDIA Volta or later) is required. This can be either on-premise or in a cloud provider, or within a managed Kubernetes environment so long as the environment has GPU support enabled.
The Helm deployment uses multiple Kubernetes secrets for obtaining access to NGC:
imagepullsecret: one for Docker images
modelpullsecret: one for model artifacts
riva-model-deploy-key: one for encypted models
The names of the secrets can be modified in the
values.yaml file, however, if you are deploying into an EGX or FleetCommander
managed environment, your environment will have support for
modelpullsecret today. These secrets
are managed by the chart, and can be manipulated by setting the respective values within the
values.yaml for Riva is intended to provide maximum flexibility in deployment configurations.
replicaCount field is used to configure the number of
identical instances (or pods) of the services that are deployed. When
load-balanced appropriately, increasing this number (as resources
permit) will enable horizontal scaling for increased load.
Individual speech services (ASR, NLP, or TTS) can be disabled by
riva.speechServices.[asr|nlp|tts] key to
Prebuilt models not required for your deployment can be deleted from
the list in
NVIDIA recommends you remove models and disable services that are not
used to reduce deployment time and GPU memory usage.
By default, models are downloaded from NGC, optimized for TensorRT (if
necessary) before the service starts, and stored in a short-lived
location. When the pod terminates, these model artifacts are deleted
and the storage is freed for other workloads. This behavior is controlled
modelDeployVolume field. See the Kubernetes Volumes
for alternative options that can be used for persistent storage. For scale
out deployments having a model store shared across pods will greatly improve
scale up time since the models will be prebuilt and already available to
the riva container.
Persistent storage should only be used in homogenous deployments where GPU models are identical.
Currently, provided models nearly fill a T4’s memory (16GB). We recommend running a subset of models/services if using a single GPU.
There is a base configuration for a simple ingress controller
using Traefik. This can be configured through the
or can be replaced with any controller supporting
Ingress controllers are found in both on-premise and cloud-based deployments. For this to work correctly, you must have a functional name resolution using whatever mechanism (DNS, /etc/host files etc).
For any sort of multi-pod scaling, you must have a correctly configured ingress controller performing http2/grpc load balancing including name resolution.
Further details can be found in the
ingress: section in the
For L2 load balancing, a barebones config using MetalLB has been supplied
and is located in the
loadbalancer: section in the
This will be useful in on-premise deployments, however, cloud-based deployments will need to use the approriate service from their provider as the networking is generally not exposed at this layer.
More details can be found in the
loadbalancer: section in the