Implementation:Bentoml BentoML DeploymentConfigParameters

Overview

DeploymentConfigParameters implements the Principle:Bentoml_BentoML_Deployment_Configuration principle by providing a structured class that captures, validates, and normalizes all deployment configuration parameters.

API

DeploymentConfigParameters class

Source

src/bentoml/_internal/cloud/deployment.py:L64-297

Import

from bentoml._internal.cloud.deployment import DeploymentConfigParameters

Signature

class DeploymentConfigParameters:
    name: str = None
    path_context: str = None
    bento: Tag | str = None
    cluster: str = None
    access_authorization: bool = None
    scaling_min: int = None
    scaling_max: int = None
    instance_type: str = None
    strategy: str = None
    labels: list[dict] = None
    envs: list[dict] = None
    secrets: list[str] = None
    extras: dict = None
    config_dict: dict = None
    config_file: str = None

Key Parameters

Parameter	Type	Default	Description
name	str	None	Deployment name (must be unique within cluster)
path_context	str	None	Path context for resolving relative bento paths
bento	str	None	Bento tag or string reference to deploy
cluster	str	None	Target cluster name
access_authorization	bool	None	Require auth for endpoint access
scaling_min	int	None	Minimum replica count (0 enables scale-to-zero)
scaling_max	int	None	Maximum replica count for auto-scaling
instance_type	str	None	Compute instance specification
strategy	str	None	Deployment update strategy
labels	list[dict]	None	Key-value metadata labels
envs	list[dict]	None	Environment variables to inject
secrets	list[str]	None	Named secrets from BentoCloud secret store
extras	dict	None	Additional provider-specific configuration
config_dict	dict	None	Programmatic configuration dictionary
config_file	str	None	Path to YAML configuration file

Inputs and Outputs

Inputs:

• Deployment parameters provided programmatically or via YAML configuration file
• Parameters can be supplied through any combination of direct arguments, config_dict, and config_file

Outputs:

• Validated DeploymentConfigParameters instance
• The verify() method validates the configuration against the target cluster's available resources

Verification

The verify() method performs validation against the cluster:

config = DeploymentConfigParameters(
    name="my-deployment",
    bento="my_service:latest",
    cluster="gcp-us-central1",
    scaling_min=1,
    scaling_max=5,
    instance_type="gpu.a10.1",
)

# Validates instance_type exists, scaling limits are valid, etc.
config.verify()

Configuration Merging

Parameters are merged with the following precedence (highest to lowest):

1. Explicit keyword arguments
2. config_dict values
3. config_file values
4. BentoCloud defaults

# config_file sets scaling_max=10, but explicit param overrides to 20
config = DeploymentConfigParameters(
    config_file="deploy.yaml",
    scaling_max=20,  # Overrides config_file value
)

YAML Configuration File Format

name: my-service
bento: my_service:v1
cluster: aws-us-east-1
access_authorization: true
scaling:
  min: 2
  max: 10
instance_type: gpu.t4.1
strategy: RollingUpdate
envs:
  - name: LOG_LEVEL
    value: INFO
secrets:
  - my-secret-group
labels:
  - key: environment
    value: production

Metadata

Property	Value
Implementation	DeploymentConfigParameters
API	DeploymentConfigParameters class
Source	src/bentoml/_internal/cloud/deployment.py:L64-297
Domain	ML_Serving, Cloud_Deployment, Infrastructure
Workflow	BentoCloud_Deployment
Principle	Principle:Bentoml_BentoML_Deployment_Configuration

Knowledge Sources

• BentoML
• BentoML Docs